---
_id: '14559'
abstract:
- lang: eng
text: We consider the problem of learning control policies in discrete-time stochastic
systems which guarantee that the system stabilizes within some specified stabilization
region with probability 1. Our approach is based on the novel notion of stabilizing
ranking supermartingales (sRSMs) that we introduce in this work. Our sRSMs overcome
the limitation of methods proposed in previous works whose applicability is restricted
to systems in which the stabilizing region cannot be left once entered under any
control policy. We present a learning procedure that learns a control policy together
with an sRSM that formally certifies probability 1 stability, both learned as
neural networks. We show that this procedure can also be adapted to formally verifying
that, under a given Lipschitz continuous control policy, the stochastic system
stabilizes within some stabilizing region with probability 1. Our experimental
evaluation shows that our learning procedure can successfully learn provably stabilizing
policies in practice.
acknowledgement: This work was supported in part by the ERC-2020-AdG 101020093, ERC
CoG 863818 (FoRM-SMArt) and the European Union’s Horizon 2020 research and innovation
programme under the Marie Skłodowska-Curie Grant Agreement No. 665385.
alternative_title:
- LNCS
article_processing_charge: No
author:
- first_name: Matin
full_name: Ansaripour, Matin
last_name: Ansaripour
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
citation:
ama: 'Ansaripour M, Chatterjee K, Henzinger TA, Lechner M, Zikelic D. Learning provably
stabilizing neural controllers for discrete-time stochastic systems. In: 21st
International Symposium on Automated Technology for Verification and Analysis.
Vol 14215. Springer Nature; 2023:357-379. doi:10.1007/978-3-031-45329-8_17'
apa: 'Ansaripour, M., Chatterjee, K., Henzinger, T. A., Lechner, M., & Zikelic,
D. (2023). Learning provably stabilizing neural controllers for discrete-time
stochastic systems. In 21st International Symposium on Automated Technology
for Verification and Analysis (Vol. 14215, pp. 357–379). Singapore, Singapore:
Springer Nature. https://doi.org/10.1007/978-3-031-45329-8_17'
chicago: Ansaripour, Matin, Krishnendu Chatterjee, Thomas A Henzinger, Mathias Lechner,
and Dorde Zikelic. “Learning Provably Stabilizing Neural Controllers for Discrete-Time
Stochastic Systems.” In 21st International Symposium on Automated Technology
for Verification and Analysis, 14215:357–79. Springer Nature, 2023. https://doi.org/10.1007/978-3-031-45329-8_17.
ieee: M. Ansaripour, K. Chatterjee, T. A. Henzinger, M. Lechner, and D. Zikelic,
“Learning provably stabilizing neural controllers for discrete-time stochastic
systems,” in 21st International Symposium on Automated Technology for Verification
and Analysis, Singapore, Singapore, 2023, vol. 14215, pp. 357–379.
ista: 'Ansaripour M, Chatterjee K, Henzinger TA, Lechner M, Zikelic D. 2023. Learning
provably stabilizing neural controllers for discrete-time stochastic systems.
21st International Symposium on Automated Technology for Verification and Analysis.
ATVA: Automated Technology for Verification and Analysis, LNCS, vol. 14215, 357–379.'
mla: Ansaripour, Matin, et al. “Learning Provably Stabilizing Neural Controllers
for Discrete-Time Stochastic Systems.” 21st International Symposium on Automated
Technology for Verification and Analysis, vol. 14215, Springer Nature, 2023,
pp. 357–79, doi:10.1007/978-3-031-45329-8_17.
short: M. Ansaripour, K. Chatterjee, T.A. Henzinger, M. Lechner, D. Zikelic, in:,
21st International Symposium on Automated Technology for Verification and Analysis,
Springer Nature, 2023, pp. 357–379.
conference:
end_date: 2023-10-27
location: Singapore, Singapore
name: 'ATVA: Automated Technology for Verification and Analysis'
start_date: 2023-10-24
date_created: 2023-11-19T23:00:56Z
date_published: 2023-10-22T00:00:00Z
date_updated: 2025-07-14T09:09:59Z
day: '22'
department:
- _id: ToHe
- _id: KrCh
doi: 10.1007/978-3-031-45329-8_17
ec_funded: 1
intvolume: ' 14215'
language:
- iso: eng
month: '10'
oa_version: None
page: 357-379
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
publication: 21st International Symposium on Automated Technology for Verification
and Analysis
publication_identifier:
eissn:
- 1611-3349
isbn:
- '9783031453281'
issn:
- 0302-9743
publication_status: published
publisher: Springer Nature
quality_controlled: '1'
scopus_import: '1'
status: public
title: Learning provably stabilizing neural controllers for discrete-time stochastic
systems
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 14215
year: '2023'
...
---
_id: '14830'
abstract:
- lang: eng
text: We study the problem of learning controllers for discrete-time non-linear
stochastic dynamical systems with formal reach-avoid guarantees. This work presents
the first method for providing formal reach-avoid guarantees, which combine and
generalize stability and safety guarantees, with a tolerable probability threshold
p in [0,1] over the infinite time horizon. Our method leverages advances in machine
learning literature and it represents formal certificates as neural networks.
In particular, we learn a certificate in the form of a reach-avoid supermartingale
(RASM), a novel notion that we introduce in this work. Our RASMs provide reachability
and avoidance guarantees by imposing constraints on what can be viewed as a stochastic
extension of level sets of Lyapunov functions for deterministic systems. Our approach
solves several important problems -- it can be used to learn a control policy
from scratch, to verify a reach-avoid specification for a fixed control policy,
or to fine-tune a pre-trained policy if it does not satisfy the reach-avoid specification.
We validate our approach on 3 stochastic non-linear reinforcement learning tasks.
acknowledgement: This work was supported in part by the ERC-2020-AdG 101020093, ERC
CoG 863818 (FoRM-SMArt) and the European Union’s Horizon 2020 research and innovation
programme under the Marie Skłodowska-Curie Grant Agreement No. 665385.
article_processing_charge: No
arxiv: 1
author:
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
citation:
ama: 'Zikelic D, Lechner M, Henzinger TA, Chatterjee K. Learning control policies
for stochastic systems with reach-avoid guarantees. In: Proceedings of the
37th AAAI Conference on Artificial Intelligence. Vol 37. Association for the
Advancement of Artificial Intelligence; 2023:11926-11935. doi:10.1609/aaai.v37i10.26407'
apa: 'Zikelic, D., Lechner, M., Henzinger, T. A., & Chatterjee, K. (2023). Learning
control policies for stochastic systems with reach-avoid guarantees. In Proceedings
of the 37th AAAI Conference on Artificial Intelligence (Vol. 37, pp. 11926–11935).
Washington, DC, United States: Association for the Advancement of Artificial Intelligence.
https://doi.org/10.1609/aaai.v37i10.26407'
chicago: Zikelic, Dorde, Mathias Lechner, Thomas A Henzinger, and Krishnendu Chatterjee.
“Learning Control Policies for Stochastic Systems with Reach-Avoid Guarantees.”
In Proceedings of the 37th AAAI Conference on Artificial Intelligence,
37:11926–35. Association for the Advancement of Artificial Intelligence, 2023.
https://doi.org/10.1609/aaai.v37i10.26407.
ieee: D. Zikelic, M. Lechner, T. A. Henzinger, and K. Chatterjee, “Learning control
policies for stochastic systems with reach-avoid guarantees,” in Proceedings
of the 37th AAAI Conference on Artificial Intelligence, Washington, DC, United
States, 2023, vol. 37, no. 10, pp. 11926–11935.
ista: 'Zikelic D, Lechner M, Henzinger TA, Chatterjee K. 2023. Learning control
policies for stochastic systems with reach-avoid guarantees. Proceedings of the
37th AAAI Conference on Artificial Intelligence. AAAI: Conference on Artificial
Intelligence vol. 37, 11926–11935.'
mla: Zikelic, Dorde, et al. “Learning Control Policies for Stochastic Systems with
Reach-Avoid Guarantees.” Proceedings of the 37th AAAI Conference on Artificial
Intelligence, vol. 37, no. 10, Association for the Advancement of Artificial
Intelligence, 2023, pp. 11926–35, doi:10.1609/aaai.v37i10.26407.
short: D. Zikelic, M. Lechner, T.A. Henzinger, K. Chatterjee, in:, Proceedings of
the 37th AAAI Conference on Artificial Intelligence, Association for the Advancement
of Artificial Intelligence, 2023, pp. 11926–11935.
conference:
end_date: 2023-02-14
location: Washington, DC, United States
name: 'AAAI: Conference on Artificial Intelligence'
start_date: 2023-02-07
date_created: 2024-01-18T07:44:31Z
date_published: 2023-06-26T00:00:00Z
date_updated: 2025-07-14T09:10:02Z
day: '26'
department:
- _id: ToHe
- _id: KrCh
doi: 10.1609/aaai.v37i10.26407
ec_funded: 1
external_id:
arxiv:
- '2210.05308'
intvolume: ' 37'
issue: '10'
keyword:
- General Medicine
language:
- iso: eng
month: '06'
oa_version: Preprint
page: 11926-11935
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
publication: Proceedings of the 37th AAAI Conference on Artificial Intelligence
publication_identifier:
eissn:
- 2374-3468
issn:
- 2159-5399
publication_status: published
publisher: Association for the Advancement of Artificial Intelligence
quality_controlled: '1'
related_material:
record:
- id: '14600'
relation: earlier_version
status: public
status: public
title: Learning control policies for stochastic systems with reach-avoid guarantees
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 37
year: '2023'
...
---
_id: '15023'
abstract:
- lang: eng
text: Reinforcement learning has shown promising results in learning neural network
policies for complicated control tasks. However, the lack of formal guarantees
about the behavior of such policies remains an impediment to their deployment.
We propose a novel method for learning a composition of neural network policies
in stochastic environments, along with a formal certificate which guarantees that
a specification over the policy's behavior is satisfied with the desired probability.
Unlike prior work on verifiable RL, our approach leverages the compositional nature
of logical specifications provided in SpectRL, to learn over graphs of probabilistic
reach-avoid specifications. The formal guarantees are provided by learning neural
network policies together with reach-avoid supermartingales (RASM) for the graph’s
sub-tasks and then composing them into a global policy. We also derive a tighter
lower bound compared to previous work on the probability of reach-avoidance implied
by a RASM, which is required to find a compositional policy with an acceptable
probabilistic threshold for complex tasks with multiple edge policies. We implement
a prototype of our approach and evaluate it on a Stochastic Nine Rooms environment.
acknowledgement: "This work was supported in part by the ERC-2020-AdG 101020093 (VAMOS)
and the ERC-2020-\r\nCoG 863818 (FoRM-SMArt)."
article_processing_charge: No
arxiv: 1
author:
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Abhinav
full_name: Verma, Abhinav
id: a235593c-d7fa-11eb-a0c5-b22ca3c66ee6
last_name: Verma
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
citation:
ama: 'Zikelic D, Lechner M, Verma A, Chatterjee K, Henzinger TA. Compositional policy
learning in stochastic control systems with formal guarantees. In: 37th Conference
on Neural Information Processing Systems. ; 2023.'
apa: Zikelic, D., Lechner, M., Verma, A., Chatterjee, K., & Henzinger, T. A.
(2023). Compositional policy learning in stochastic control systems with formal
guarantees. In 37th Conference on Neural Information Processing Systems.
New Orleans, LO, United States.
chicago: Zikelic, Dorde, Mathias Lechner, Abhinav Verma, Krishnendu Chatterjee,
and Thomas A Henzinger. “Compositional Policy Learning in Stochastic Control Systems
with Formal Guarantees.” In 37th Conference on Neural Information Processing
Systems, 2023.
ieee: D. Zikelic, M. Lechner, A. Verma, K. Chatterjee, and T. A. Henzinger, “Compositional
policy learning in stochastic control systems with formal guarantees,” in 37th
Conference on Neural Information Processing Systems, New Orleans, LO, United
States, 2023.
ista: 'Zikelic D, Lechner M, Verma A, Chatterjee K, Henzinger TA. 2023. Compositional
policy learning in stochastic control systems with formal guarantees. 37th Conference
on Neural Information Processing Systems. NeurIPS: Neural Information Processing
Systems.'
mla: Zikelic, Dorde, et al. “Compositional Policy Learning in Stochastic Control
Systems with Formal Guarantees.” 37th Conference on Neural Information Processing
Systems, 2023.
short: D. Zikelic, M. Lechner, A. Verma, K. Chatterjee, T.A. Henzinger, in:, 37th
Conference on Neural Information Processing Systems, 2023.
conference:
end_date: 2023-12-16
location: New Orleans, LO, United States
name: 'NeurIPS: Neural Information Processing Systems'
start_date: 2023-12-10
date_created: 2024-02-25T09:23:24Z
date_published: 2023-12-15T00:00:00Z
date_updated: 2025-07-14T09:10:04Z
day: '15'
department:
- _id: ToHe
- _id: KrCh
ec_funded: 1
external_id:
arxiv:
- '2312.01456'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://doi.org/10.48550/arXiv.2312.01456
month: '12'
oa: 1
oa_version: Preprint
project:
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
publication: 37th Conference on Neural Information Processing Systems
publication_status: epub_ahead
quality_controlled: '1'
status: public
title: Compositional policy learning in stochastic control systems with formal guarantees
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
year: '2023'
...
---
_id: '13142'
abstract:
- lang: eng
text: Reinforcement learning has received much attention for learning controllers
of deterministic systems. We consider a learner-verifier framework for stochastic
control systems and survey recent methods that formally guarantee a conjunction
of reachability and safety properties. Given a property and a lower bound on the
probability of the property being satisfied, our framework jointly learns a control
policy and a formal certificate to ensure the satisfaction of the property with
a desired probability threshold. Both the control policy and the formal certificate
are continuous functions from states to reals, which are learned as parameterized
neural networks. While in the deterministic case, the certificates are invariant
and barrier functions for safety, or Lyapunov and ranking functions for liveness,
in the stochastic case the certificates are supermartingales. For certificate
verification, we use interval arithmetic abstract interpretation to bound the
expected values of neural network functions.
acknowledgement: This work was supported in part by the ERC-2020-AdG 101020093, ERC
CoG 863818 (FoRM-SMArt) and the European Union’s Horizon 2020 research and innovation
programme under the Marie Skłodowska-Curie Grant Agreement No. 665385.
alternative_title:
- LNCS
article_processing_charge: No
author:
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
citation:
ama: 'Chatterjee K, Henzinger TA, Lechner M, Zikelic D. A learner-verifier framework
for neural network controllers and certificates of stochastic systems. In: Tools
and Algorithms for the Construction and Analysis of Systems . Vol 13993. Springer
Nature; 2023:3-25. doi:10.1007/978-3-031-30823-9_1'
apa: 'Chatterjee, K., Henzinger, T. A., Lechner, M., & Zikelic, D. (2023). A
learner-verifier framework for neural network controllers and certificates of
stochastic systems. In Tools and Algorithms for the Construction and Analysis
of Systems (Vol. 13993, pp. 3–25). Paris, France: Springer Nature. https://doi.org/10.1007/978-3-031-30823-9_1'
chicago: Chatterjee, Krishnendu, Thomas A Henzinger, Mathias Lechner, and Dorde
Zikelic. “A Learner-Verifier Framework for Neural Network Controllers and Certificates
of Stochastic Systems.” In Tools and Algorithms for the Construction and Analysis
of Systems , 13993:3–25. Springer Nature, 2023. https://doi.org/10.1007/978-3-031-30823-9_1.
ieee: K. Chatterjee, T. A. Henzinger, M. Lechner, and D. Zikelic, “A learner-verifier
framework for neural network controllers and certificates of stochastic systems,”
in Tools and Algorithms for the Construction and Analysis of Systems ,
Paris, France, 2023, vol. 13993, pp. 3–25.
ista: 'Chatterjee K, Henzinger TA, Lechner M, Zikelic D. 2023. A learner-verifier
framework for neural network controllers and certificates of stochastic systems.
Tools and Algorithms for the Construction and Analysis of Systems . TACAS: Tools
and Algorithms for the Construction and Analysis of Systems, LNCS, vol. 13993,
3–25.'
mla: Chatterjee, Krishnendu, et al. “A Learner-Verifier Framework for Neural Network
Controllers and Certificates of Stochastic Systems.” Tools and Algorithms for
the Construction and Analysis of Systems , vol. 13993, Springer Nature, 2023,
pp. 3–25, doi:10.1007/978-3-031-30823-9_1.
short: K. Chatterjee, T.A. Henzinger, M. Lechner, D. Zikelic, in:, Tools and Algorithms
for the Construction and Analysis of Systems , Springer Nature, 2023, pp. 3–25.
conference:
end_date: 2023-04-27
location: Paris, France
name: 'TACAS: Tools and Algorithms for the Construction and Analysis of Systems'
start_date: 2023-04-22
date_created: 2023-06-18T22:00:47Z
date_published: 2023-04-22T00:00:00Z
date_updated: 2025-07-14T09:09:52Z
day: '22'
ddc:
- '000'
department:
- _id: KrCh
- _id: ToHe
doi: 10.1007/978-3-031-30823-9_1
ec_funded: 1
file:
- access_level: open_access
checksum: 3d8a8bb24d211bc83360dfc2fd744307
content_type: application/pdf
creator: dernst
date_created: 2023-06-19T08:29:30Z
date_updated: 2023-06-19T08:29:30Z
file_id: '13150'
file_name: 2023_LNCS_Chatterjee.pdf
file_size: 528455
relation: main_file
success: 1
file_date_updated: 2023-06-19T08:29:30Z
has_accepted_license: '1'
intvolume: ' 13993'
language:
- iso: eng
month: '04'
oa: 1
oa_version: Published Version
page: 3-25
project:
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
publication: 'Tools and Algorithms for the Construction and Analysis of Systems '
publication_identifier:
eissn:
- 1611-3349
isbn:
- '9783031308222'
issn:
- 0302-9743
publication_status: published
publisher: Springer Nature
quality_controlled: '1'
scopus_import: '1'
status: public
title: A learner-verifier framework for neural network controllers and certificates
of stochastic systems
tmp:
image: /images/cc_by.png
legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode
name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)
short: CC BY (4.0)
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 13993
year: '2023'
...
---
_id: '14242'
abstract:
- lang: eng
text: We study the problem of training and certifying adversarially robust quantized
neural networks (QNNs). Quantization is a technique for making neural networks
more efficient by running them using low-bit integer arithmetic and is therefore
commonly adopted in industry. Recent work has shown that floating-point neural
networks that have been verified to be robust can become vulnerable to adversarial
attacks after quantization, and certification of the quantized representation
is necessary to guarantee robustness. In this work, we present quantization-aware
interval bound propagation (QA-IBP), a novel method for training robust QNNs.
Inspired by advances in robust learning of non-quantized networks, our training
algorithm computes the gradient of an abstract representation of the actual network.
Unlike existing approaches, our method can handle the discrete semantics of QNNs.
Based on QA-IBP, we also develop a complete verification procedure for verifying
the adversarial robustness of QNNs, which is guaranteed to terminate and produce
a correct answer. Compared to existing approaches, the key advantage of our verification
procedure is that it runs entirely on GPU or other accelerator devices. We demonstrate
experimentally that our approach significantly outperforms existing methods and
establish the new state-of-the-art for training and certifying the robustness
of QNNs.
acknowledgement: "This work was supported in part by the ERC-2020-AdG 101020093, ERC
CoG 863818 (FoRM-SMArt) and the European Union’s Horizon 2020 research and innovation
programme under the Marie Skłodowska-Curie Grant Agreement No. 665385. Research
was sponsored by the United\r\nStates Air Force Research Laboratory and the United
States Air Force Artificial Intelligence Accelerator and was accomplished under
Cooperative Agreement Number FA8750-19-2-\r\n1000. The views and conclusions contained
in this document are those of the authors and should not be interpreted as representing
the official policies, either expressed or implied,\r\nof the United States Air
Force or the U.S. Government. The U.S. Government is authorized to reproduce and
distribute reprints for Government purposes notwithstanding any copyright\r\nnotation
herein. The research was also funded in part by the AI2050 program at Schmidt Futures
(Grant G-22-63172) and Capgemini SE."
article_processing_charge: No
arxiv: 1
author:
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
citation:
ama: 'Lechner M, Zikelic D, Chatterjee K, Henzinger TA, Rus D. Quantization-aware
interval bound propagation for training certifiably robust quantized neural networks.
In: Proceedings of the 37th AAAI Conference on Artificial Intelligence.
Vol 37. Association for the Advancement of Artificial Intelligence; 2023:14964-14973.
doi:10.1609/aaai.v37i12.26747'
apa: 'Lechner, M., Zikelic, D., Chatterjee, K., Henzinger, T. A., & Rus, D.
(2023). Quantization-aware interval bound propagation for training certifiably
robust quantized neural networks. In Proceedings of the 37th AAAI Conference
on Artificial Intelligence (Vol. 37, pp. 14964–14973). Washington, DC, United
States: Association for the Advancement of Artificial Intelligence. https://doi.org/10.1609/aaai.v37i12.26747'
chicago: Lechner, Mathias, Dorde Zikelic, Krishnendu Chatterjee, Thomas A Henzinger,
and Daniela Rus. “Quantization-Aware Interval Bound Propagation for Training Certifiably
Robust Quantized Neural Networks.” In Proceedings of the 37th AAAI Conference
on Artificial Intelligence, 37:14964–73. Association for the Advancement of
Artificial Intelligence, 2023. https://doi.org/10.1609/aaai.v37i12.26747.
ieee: M. Lechner, D. Zikelic, K. Chatterjee, T. A. Henzinger, and D. Rus, “Quantization-aware
interval bound propagation for training certifiably robust quantized neural networks,”
in Proceedings of the 37th AAAI Conference on Artificial Intelligence,
Washington, DC, United States, 2023, vol. 37, no. 12, pp. 14964–14973.
ista: 'Lechner M, Zikelic D, Chatterjee K, Henzinger TA, Rus D. 2023. Quantization-aware
interval bound propagation for training certifiably robust quantized neural networks.
Proceedings of the 37th AAAI Conference on Artificial Intelligence. AAAI: Conference
on Artificial Intelligence vol. 37, 14964–14973.'
mla: Lechner, Mathias, et al. “Quantization-Aware Interval Bound Propagation for
Training Certifiably Robust Quantized Neural Networks.” Proceedings of the
37th AAAI Conference on Artificial Intelligence, vol. 37, no. 12, Association
for the Advancement of Artificial Intelligence, 2023, pp. 14964–73, doi:10.1609/aaai.v37i12.26747.
short: M. Lechner, D. Zikelic, K. Chatterjee, T.A. Henzinger, D. Rus, in:, Proceedings
of the 37th AAAI Conference on Artificial Intelligence, Association for the Advancement
of Artificial Intelligence, 2023, pp. 14964–14973.
conference:
end_date: 2023-02-14
location: Washington, DC, United States
name: 'AAAI: Conference on Artificial Intelligence'
start_date: 2023-02-07
date_created: 2023-08-27T22:01:17Z
date_published: 2023-06-26T00:00:00Z
date_updated: 2025-07-14T09:09:56Z
day: '26'
department:
- _id: ToHe
- _id: KrCh
doi: 10.1609/aaai.v37i12.26747
ec_funded: 1
external_id:
arxiv:
- '2211.16187'
intvolume: ' 37'
issue: '12'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://doi.org/10.48550/arXiv.2211.16187
month: '06'
oa: 1
oa_version: Preprint
page: 14964-14973
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
publication: Proceedings of the 37th AAAI Conference on Artificial Intelligence
publication_identifier:
isbn:
- '9781577358800'
publication_status: published
publisher: Association for the Advancement of Artificial Intelligence
quality_controlled: '1'
scopus_import: '1'
status: public
title: Quantization-aware interval bound propagation for training certifiably robust
quantized neural networks
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 37
year: '2023'
...
---
_id: '12704'
abstract:
- lang: eng
text: Adversarial training (i.e., training on adversarially perturbed input data)
is a well-studied method for making neural networks robust to potential adversarial
attacks during inference. However, the improved robustness does not come for free
but rather is accompanied by a decrease in overall model accuracy and performance.
Recent work has shown that, in practical robot learning applications, the effects
of adversarial training do not pose a fair trade-off but inflict a net loss when
measured in holistic robot performance. This work revisits the robustness-accuracy
trade-off in robot learning by systematically analyzing if recent advances in
robust training methods and theory in conjunction with adversarial robot learning,
are capable of making adversarial training suitable for real-world robot applications.
We evaluate three different robot learning tasks ranging from autonomous driving
in a high-fidelity environment amenable to sim-to-real deployment to mobile robot
navigation and gesture recognition. Our results demonstrate that, while these
techniques make incremental improvements on the trade-off on a relative scale,
the negative impact on the nominal accuracy caused by adversarial training still
outweighs the improved robustness by an order of magnitude. We conclude that although
progress is happening, further advances in robust learning methods are necessary
before they can benefit robot learning tasks in practice.
acknowledgement: "We thank Christoph Lampert for inspiring this work. The\r\nviews
and conclusions contained in this document are those of\r\nthe authors and should
not be interpreted as representing the\r\nofficial policies, either expressed or
implied, of the United States\r\nAir Force or the U.S. Government. The U.S. Government
is\r\nauthorized to reproduce and distribute reprints for Government\r\npurposes
notwithstanding any copyright notation herein."
article_processing_charge: No
article_type: original
arxiv: 1
author:
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Alexander
full_name: Amini, Alexander
last_name: Amini
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
citation:
ama: Lechner M, Amini A, Rus D, Henzinger TA. Revisiting the adversarial robustness-accuracy
tradeoff in robot learning. IEEE Robotics and Automation Letters. 2023;8(3):1595-1602.
doi:10.1109/LRA.2023.3240930
apa: Lechner, M., Amini, A., Rus, D., & Henzinger, T. A. (2023). Revisiting
the adversarial robustness-accuracy tradeoff in robot learning. IEEE Robotics
and Automation Letters. Institute of Electrical and Electronics Engineers.
https://doi.org/10.1109/LRA.2023.3240930
chicago: Lechner, Mathias, Alexander Amini, Daniela Rus, and Thomas A Henzinger.
“Revisiting the Adversarial Robustness-Accuracy Tradeoff in Robot Learning.” IEEE
Robotics and Automation Letters. Institute of Electrical and Electronics Engineers,
2023. https://doi.org/10.1109/LRA.2023.3240930.
ieee: M. Lechner, A. Amini, D. Rus, and T. A. Henzinger, “Revisiting the adversarial
robustness-accuracy tradeoff in robot learning,” IEEE Robotics and Automation
Letters, vol. 8, no. 3. Institute of Electrical and Electronics Engineers,
pp. 1595–1602, 2023.
ista: Lechner M, Amini A, Rus D, Henzinger TA. 2023. Revisiting the adversarial
robustness-accuracy tradeoff in robot learning. IEEE Robotics and Automation Letters.
8(3), 1595–1602.
mla: Lechner, Mathias, et al. “Revisiting the Adversarial Robustness-Accuracy Tradeoff
in Robot Learning.” IEEE Robotics and Automation Letters, vol. 8, no. 3,
Institute of Electrical and Electronics Engineers, 2023, pp. 1595–602, doi:10.1109/LRA.2023.3240930.
short: M. Lechner, A. Amini, D. Rus, T.A. Henzinger, IEEE Robotics and Automation
Letters 8 (2023) 1595–1602.
date_created: 2023-03-05T23:01:04Z
date_published: 2023-03-01T00:00:00Z
date_updated: 2023-08-01T13:36:50Z
day: '01'
ddc:
- '000'
department:
- _id: ToHe
doi: 10.1109/LRA.2023.3240930
external_id:
arxiv:
- '2204.07373'
isi:
- '000936534100012'
file:
- access_level: open_access
checksum: 5a75dcd326ea66685de2b1aaec259e85
content_type: application/pdf
creator: cchlebak
date_created: 2023-03-07T12:22:23Z
date_updated: 2023-03-07T12:22:23Z
file_id: '12714'
file_name: 2023_IEEERobAutLetters_Lechner.pdf
file_size: 944052
relation: main_file
success: 1
file_date_updated: 2023-03-07T12:22:23Z
has_accepted_license: '1'
intvolume: ' 8'
isi: 1
issue: '3'
language:
- iso: eng
month: '03'
oa: 1
oa_version: Published Version
page: 1595-1602
publication: IEEE Robotics and Automation Letters
publication_identifier:
eissn:
- 2377-3766
publication_status: published
publisher: Institute of Electrical and Electronics Engineers
quality_controlled: '1'
related_material:
record:
- id: '11366'
relation: earlier_version
status: public
scopus_import: '1'
status: public
title: Revisiting the adversarial robustness-accuracy tradeoff in robot learning
tmp:
image: /images/cc_by.png
legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode
name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)
short: CC BY (4.0)
type: journal_article
user_id: 4359f0d1-fa6c-11eb-b949-802e58b17ae8
volume: 8
year: '2023'
...
---
_id: '11362'
abstract:
- lang: eng
text: "Deep learning has enabled breakthroughs in challenging computing problems
and has emerged as the standard problem-solving tool for computer vision and natural
language processing tasks.\r\nOne exception to this trend is safety-critical tasks
where robustness and resilience requirements contradict the black-box nature of
neural networks. \r\nTo deploy deep learning methods for these tasks, it is vital
to provide guarantees on neural network agents' safety and robustness criteria.
\r\nThis can be achieved by developing formal verification methods to verify the
safety and robustness properties of neural networks.\r\n\r\nOur goal is to design,
develop and assess safety verification methods for neural networks to improve
their reliability and trustworthiness in real-world applications.\r\nThis thesis
establishes techniques for the verification of compressed and adversarially trained
models as well as the design of novel neural networks for verifiably safe decision-making.\r\n\r\nFirst,
we establish the problem of verifying quantized neural networks. Quantization
is a technique that trades numerical precision for the computational efficiency
of running a neural network and is widely adopted in industry.\r\nWe show that
neglecting the reduced precision when verifying a neural network can lead to wrong
conclusions about the robustness and safety of the network, highlighting that
novel techniques for quantized network verification are necessary. We introduce
several bit-exact verification methods explicitly designed for quantized neural
networks and experimentally confirm on realistic networks that the network's robustness
and other formal properties are affected by the quantization.\r\n\r\nFurthermore,
we perform a case study providing evidence that adversarial training, a standard
technique for making neural networks more robust, has detrimental effects on the
network's performance. This robustness-accuracy tradeoff has been studied before
regarding the accuracy obtained on classification datasets where each data point
is independent of all other data points. On the other hand, we investigate the
tradeoff empirically in robot learning settings where a both, a high accuracy
and a high robustness, are desirable.\r\nOur results suggest that the negative
side-effects of adversarial training outweigh its robustness benefits in practice.\r\n\r\nFinally,
we consider the problem of verifying safety when running a Bayesian neural network
policy in a feedback loop with systems over the infinite time horizon. Bayesian
neural networks are probabilistic models for learning uncertainties in the data
and are therefore often used on robotic and healthcare applications where data
is inherently stochastic.\r\nWe introduce a method for recalibrating Bayesian
neural networks so that they yield probability distributions over safe decisions
only.\r\nOur method learns a safety certificate that guarantees safety over the
infinite time horizon to determine which decisions are safe in every possible
state of the system.\r\nWe demonstrate the effectiveness of our approach on a
series of reinforcement learning benchmarks."
alternative_title:
- ISTA Thesis
article_processing_charge: No
author:
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
citation:
ama: Lechner M. Learning verifiable representations. 2022. doi:10.15479/at:ista:11362
apa: Lechner, M. (2022). Learning verifiable representations. Institute of
Science and Technology Austria. https://doi.org/10.15479/at:ista:11362
chicago: Lechner, Mathias. “Learning Verifiable Representations.” Institute of Science
and Technology Austria, 2022. https://doi.org/10.15479/at:ista:11362.
ieee: M. Lechner, “Learning verifiable representations,” Institute of Science and
Technology Austria, 2022.
ista: Lechner M. 2022. Learning verifiable representations. Institute of Science
and Technology Austria.
mla: Lechner, Mathias. Learning Verifiable Representations. Institute of
Science and Technology Austria, 2022, doi:10.15479/at:ista:11362.
short: M. Lechner, Learning Verifiable Representations, Institute of Science and
Technology Austria, 2022.
date_created: 2022-05-12T07:14:01Z
date_published: 2022-05-12T00:00:00Z
date_updated: 2025-07-14T09:10:11Z
day: '12'
ddc:
- '004'
degree_awarded: PhD
department:
- _id: GradSch
- _id: ToHe
doi: 10.15479/at:ista:11362
ec_funded: 1
file:
- access_level: closed
checksum: 8eefa9c7c10ca7e1a2ccdd731962a645
content_type: application/zip
creator: mlechner
date_created: 2022-05-13T12:33:26Z
date_updated: 2022-05-13T12:49:00Z
file_id: '11378'
file_name: src.zip
file_size: 13210143
relation: source_file
- access_level: open_access
checksum: 1b9e1e5a9a83ed9d89dad2f5133dc026
content_type: application/pdf
creator: mlechner
date_created: 2022-05-16T08:02:28Z
date_updated: 2022-05-17T15:19:39Z
file_id: '11382'
file_name: thesis_main-a2.pdf
file_size: 2732536
relation: main_file
file_date_updated: 2022-05-17T15:19:39Z
has_accepted_license: '1'
keyword:
- neural networks
- verification
- machine learning
language:
- iso: eng
month: '05'
oa: 1
oa_version: Published Version
page: '124'
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
publication_identifier:
isbn:
- 978-3-99078-017-6
publication_status: published
publisher: Institute of Science and Technology Austria
related_material:
record:
- id: '11366'
relation: part_of_dissertation
status: public
- id: '7808'
relation: part_of_dissertation
status: public
- id: '10666'
relation: part_of_dissertation
status: public
- id: '10665'
relation: part_of_dissertation
status: public
- id: '10667'
relation: part_of_dissertation
status: public
status: public
supervisor:
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
title: Learning verifiable representations
tmp:
image: /image/cc_by_nd.png
legal_code_url: https://creativecommons.org/licenses/by-nd/4.0/legalcode
name: Creative Commons Attribution-NoDerivatives 4.0 International (CC BY-ND 4.0)
short: CC BY-ND (4.0)
type: dissertation
user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9
year: '2022'
...
---
_id: '11366'
abstract:
- lang: eng
text: "Adversarial training (i.e., training on adversarially perturbed input data)
is a well-studied method for making neural networks robust to potential adversarial
attacks during inference. However, the improved robustness does not\r\ncome for
free but rather is accompanied by a decrease in overall model accuracy and performance.
Recent work has shown that, in practical robot learning applications, the effects
of adversarial training do not pose a fair trade-off\r\nbut inflict a net loss
when measured in holistic robot performance. This work revisits the robustness-accuracy
trade-off in robot learning by systematically analyzing if recent advances in
robust training methods and theory in\r\nconjunction with adversarial robot learning
can make adversarial training suitable for real-world robot applications. We evaluate
a wide variety of robot learning tasks ranging from autonomous driving in a high-fidelity
environment\r\namenable to sim-to-real deployment, to mobile robot gesture recognition.
Our results demonstrate that, while these techniques make incremental improvements
on the trade-off on a relative scale, the negative side-effects caused by\r\nadversarial
training still outweigh the improvements by an order of magnitude. We conclude
that more substantial advances in robust learning methods are necessary before
they can benefit robot learning tasks in practice."
acknowledgement: "This work was supported in parts by the ERC-2020-AdG 101020093,
National Science Foundation (NSF), and JP\r\nMorgan Graduate Fellowships. We thank
Christoph Lampert for inspiring this work.\r\n"
article_number: '2204.07373'
article_processing_charge: No
arxiv: 1
author:
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Alexander
full_name: Amini, Alexander
last_name: Amini
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
citation:
ama: Lechner M, Amini A, Rus D, Henzinger TA. Revisiting the adversarial robustness-accuracy
tradeoff in robot learning. arXiv. doi:10.48550/arXiv.2204.07373
apa: Lechner, M., Amini, A., Rus, D., & Henzinger, T. A. (n.d.). Revisiting
the adversarial robustness-accuracy tradeoff in robot learning. arXiv.
https://doi.org/10.48550/arXiv.2204.07373
chicago: Lechner, Mathias, Alexander Amini, Daniela Rus, and Thomas A Henzinger.
“Revisiting the Adversarial Robustness-Accuracy Tradeoff in Robot Learning.” ArXiv,
n.d. https://doi.org/10.48550/arXiv.2204.07373.
ieee: M. Lechner, A. Amini, D. Rus, and T. A. Henzinger, “Revisiting the adversarial
robustness-accuracy tradeoff in robot learning,” arXiv. .
ista: Lechner M, Amini A, Rus D, Henzinger TA. Revisiting the adversarial robustness-accuracy
tradeoff in robot learning. arXiv, 2204.07373.
mla: Lechner, Mathias, et al. “Revisiting the Adversarial Robustness-Accuracy Tradeoff
in Robot Learning.” ArXiv, 2204.07373, doi:10.48550/arXiv.2204.07373.
short: M. Lechner, A. Amini, D. Rus, T.A. Henzinger, ArXiv (n.d.).
date_created: 2022-05-12T13:20:17Z
date_published: 2022-04-15T00:00:00Z
date_updated: 2023-08-01T13:36:50Z
day: '15'
department:
- _id: ToHe
doi: 10.48550/arXiv.2204.07373
ec_funded: 1
external_id:
arxiv:
- '2204.07373'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://doi.org/10.48550/arXiv.2204.07373
month: '04'
oa: 1
oa_version: Preprint
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
publication: arXiv
publication_status: submitted
related_material:
record:
- id: '11362'
relation: dissertation_contains
status: public
- id: '12704'
relation: later_version
status: public
status: public
title: Revisiting the adversarial robustness-accuracy tradeoff in robot learning
type: preprint
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
year: '2022'
...
---
_id: '14600'
abstract:
- lang: eng
text: We study the problem of learning controllers for discrete-time non-linear
stochastic dynamical systems with formal reach-avoid guarantees. This work presents
the first method for providing formal reach-avoid guarantees, which combine and
generalize stability and safety guarantees, with a tolerable probability threshold
$p\in[0,1]$ over the infinite time horizon. Our method leverages advances in machine
learning literature and it represents formal certificates as neural networks.
In particular, we learn a certificate in the form of a reach-avoid supermartingale
(RASM), a novel notion that we introduce in this work. Our RASMs provide reachability
and avoidance guarantees by imposing constraints on what can be viewed as a stochastic
extension of level sets of Lyapunov functions for deterministic systems. Our approach
solves several important problems -- it can be used to learn a control policy
from scratch, to verify a reach-avoid specification for a fixed control policy,
or to fine-tune a pre-trained policy if it does not satisfy the reach-avoid specification.
We validate our approach on $3$ stochastic non-linear reinforcement learning tasks.
article_processing_charge: No
arxiv: 1
author:
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
citation:
ama: Zikelic D, Lechner M, Henzinger TA, Chatterjee K. Learning control policies
for stochastic systems with reach-avoid guarantees. arXiv. doi:10.48550/ARXIV.2210.05308
apa: Zikelic, D., Lechner, M., Henzinger, T. A., & Chatterjee, K. (n.d.). Learning
control policies for stochastic systems with reach-avoid guarantees. arXiv.
https://doi.org/10.48550/ARXIV.2210.05308
chicago: Zikelic, Dorde, Mathias Lechner, Thomas A Henzinger, and Krishnendu Chatterjee.
“Learning Control Policies for Stochastic Systems with Reach-Avoid Guarantees.”
ArXiv, n.d. https://doi.org/10.48550/ARXIV.2210.05308.
ieee: D. Zikelic, M. Lechner, T. A. Henzinger, and K. Chatterjee, “Learning control
policies for stochastic systems with reach-avoid guarantees,” arXiv. .
ista: Zikelic D, Lechner M, Henzinger TA, Chatterjee K. Learning control policies
for stochastic systems with reach-avoid guarantees. arXiv, 10.48550/ARXIV.2210.05308.
mla: Zikelic, Dorde, et al. “Learning Control Policies for Stochastic Systems with
Reach-Avoid Guarantees.” ArXiv, doi:10.48550/ARXIV.2210.05308.
short: D. Zikelic, M. Lechner, T.A. Henzinger, K. Chatterjee, ArXiv (n.d.).
date_created: 2023-11-24T13:10:09Z
date_published: 2022-11-29T00:00:00Z
date_updated: 2025-07-14T09:10:02Z
day: '29'
department:
- _id: KrCh
- _id: ToHe
doi: 10.48550/ARXIV.2210.05308
ec_funded: 1
external_id:
arxiv:
- '2210.05308'
language:
- iso: eng
license: https://creativecommons.org/licenses/by-sa/4.0/
main_file_link:
- open_access: '1'
url: https://arxiv.org/abs/2210.05308
month: '11'
oa: 1
oa_version: Preprint
project:
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
publication: arXiv
publication_status: submitted
related_material:
record:
- id: '14539'
relation: dissertation_contains
status: public
- id: '14830'
relation: later_version
status: public
status: public
title: Learning control policies for stochastic systems with reach-avoid guarantees
tmp:
image: /images/cc_by_sa.png
legal_code_url: https://creativecommons.org/licenses/by-sa/4.0/legalcode
name: Creative Commons Attribution-ShareAlike 4.0 International Public License (CC
BY-SA 4.0)
short: CC BY-SA (4.0)
type: preprint
user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9
year: '2022'
...
---
_id: '14601'
abstract:
- lang: eng
text: "In this work, we address the problem of learning provably stable neural\r\nnetwork
policies for stochastic control systems. While recent work has\r\ndemonstrated
the feasibility of certifying given policies using martingale\r\ntheory, the problem
of how to learn such policies is little explored. Here, we\r\nstudy the effectiveness
of jointly learning a policy together with a martingale\r\ncertificate that proves
its stability using a single learning algorithm. We\r\nobserve that the joint
optimization problem becomes easily stuck in local\r\nminima when starting from
a randomly initialized policy. Our results suggest\r\nthat some form of pre-training
of the policy is required for the joint\r\noptimization to repair and verify the
policy successfully."
article_processing_charge: No
arxiv: 1
author:
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
citation:
ama: Zikelic D, Lechner M, Chatterjee K, Henzinger TA. Learning stabilizing policies
in stochastic control systems. arXiv. doi:10.48550/arXiv.2205.11991
apa: Zikelic, D., Lechner, M., Chatterjee, K., & Henzinger, T. A. (n.d.). Learning
stabilizing policies in stochastic control systems. arXiv. https://doi.org/10.48550/arXiv.2205.11991
chicago: Zikelic, Dorde, Mathias Lechner, Krishnendu Chatterjee, and Thomas A Henzinger.
“Learning Stabilizing Policies in Stochastic Control Systems.” ArXiv, n.d.
https://doi.org/10.48550/arXiv.2205.11991.
ieee: D. Zikelic, M. Lechner, K. Chatterjee, and T. A. Henzinger, “Learning stabilizing
policies in stochastic control systems,” arXiv. .
ista: Zikelic D, Lechner M, Chatterjee K, Henzinger TA. Learning stabilizing policies
in stochastic control systems. arXiv, 10.48550/arXiv.2205.11991.
mla: Zikelic, Dorde, et al. “Learning Stabilizing Policies in Stochastic Control
Systems.” ArXiv, doi:10.48550/arXiv.2205.11991.
short: D. Zikelic, M. Lechner, K. Chatterjee, T.A. Henzinger, ArXiv (n.d.).
date_created: 2023-11-24T13:22:30Z
date_published: 2022-05-24T00:00:00Z
date_updated: 2025-07-14T09:10:00Z
day: '24'
department:
- _id: KrCh
- _id: ToHe
doi: 10.48550/arXiv.2205.11991
ec_funded: 1
external_id:
arxiv:
- '2205.11991'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://arxiv.org/abs/2205.11991
month: '05'
oa: 1
oa_version: Preprint
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
publication: arXiv
publication_status: submitted
related_material:
record:
- id: '14539'
relation: dissertation_contains
status: public
status: public
title: Learning stabilizing policies in stochastic control systems
type: preprint
user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9
year: '2022'
...
---
_id: '12010'
abstract:
- lang: eng
text: World models learn behaviors in a latent imagination space to enhance the
sample-efficiency of deep reinforcement learning (RL) algorithms. While learning
world models for high-dimensional observations (e.g., pixel inputs) has become
practicable on standard RL benchmarks and some games, their effectiveness in real-world
robotics applications has not been explored. In this paper, we investigate how
such agents generalize to real-world autonomous vehicle control tasks, where advanced
model-free deep RL algorithms fail. In particular, we set up a series of time-lap
tasks for an F1TENTH racing robot, equipped with a high-dimensional LiDAR sensor,
on a set of test tracks with a gradual increase in their complexity. In this continuous-control
setting, we show that model-based agents capable of learning in imagination substantially
outperform model-free agents with respect to performance, sample efficiency, successful
task completion, and generalization. Moreover, we show that the generalization
ability of model-based agents strongly depends on the choice of their observation
model. We provide extensive empirical evidence for the effectiveness of world
models provided with long enough memory horizons in sim2real tasks.
acknowledgement: L.B. was supported by the Doctoral College Resilient Embedded Systems.
M.L. was supported in part by the ERC2020-AdG 101020093 and the Austrian Science
Fund (FWF) under grant Z211-N23 (Wittgenstein Award). R.H. and D.R. were supported
by The Boeing Company and the Office of Naval Research (ONR) Grant N00014-18-1-2830.
R.G. was partially supported by the Horizon-2020 ECSEL Project grant No. 783163
(iDev40) and A.B. by FFG Project ADEX.
article_processing_charge: No
arxiv: 1
author:
- first_name: Axel
full_name: Brunnbauer, Axel
last_name: Brunnbauer
- first_name: Luigi
full_name: Berducci, Luigi
last_name: Berducci
- first_name: Andreas
full_name: Brandstatter, Andreas
last_name: Brandstatter
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Ramin
full_name: Hasani, Ramin
last_name: Hasani
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
- first_name: Radu
full_name: Grosu, Radu
last_name: Grosu
citation:
ama: 'Brunnbauer A, Berducci L, Brandstatter A, et al. Latent imagination facilitates
zero-shot transfer in autonomous racing. In: 2022 International Conference
on Robotics and Automation. IEEE; 2022:7513-7520. doi:10.1109/ICRA46639.2022.9811650'
apa: 'Brunnbauer, A., Berducci, L., Brandstatter, A., Lechner, M., Hasani, R., Rus,
D., & Grosu, R. (2022). Latent imagination facilitates zero-shot transfer
in autonomous racing. In 2022 International Conference on Robotics and Automation
(pp. 7513–7520). Philadelphia, PA, United States: IEEE. https://doi.org/10.1109/ICRA46639.2022.9811650'
chicago: Brunnbauer, Axel, Luigi Berducci, Andreas Brandstatter, Mathias Lechner,
Ramin Hasani, Daniela Rus, and Radu Grosu. “Latent Imagination Facilitates Zero-Shot
Transfer in Autonomous Racing.” In 2022 International Conference on Robotics
and Automation, 7513–20. IEEE, 2022. https://doi.org/10.1109/ICRA46639.2022.9811650.
ieee: A. Brunnbauer et al., “Latent imagination facilitates zero-shot transfer
in autonomous racing,” in 2022 International Conference on Robotics and Automation,
Philadelphia, PA, United States, 2022, pp. 7513–7520.
ista: 'Brunnbauer A, Berducci L, Brandstatter A, Lechner M, Hasani R, Rus D, Grosu
R. 2022. Latent imagination facilitates zero-shot transfer in autonomous racing.
2022 International Conference on Robotics and Automation. ICRA: International
Conference on Robotics and Automation, 7513–7520.'
mla: Brunnbauer, Axel, et al. “Latent Imagination Facilitates Zero-Shot Transfer
in Autonomous Racing.” 2022 International Conference on Robotics and Automation,
IEEE, 2022, pp. 7513–20, doi:10.1109/ICRA46639.2022.9811650.
short: A. Brunnbauer, L. Berducci, A. Brandstatter, M. Lechner, R. Hasani, D. Rus,
R. Grosu, in:, 2022 International Conference on Robotics and Automation, IEEE,
2022, pp. 7513–7520.
conference:
end_date: 2022-05-27
location: Philadelphia, PA, United States
name: 'ICRA: International Conference on Robotics and Automation'
start_date: 2022-05-23
date_created: 2022-09-04T22:02:02Z
date_published: 2022-07-12T00:00:00Z
date_updated: 2022-09-05T08:46:12Z
day: '12'
department:
- _id: ToHe
doi: 10.1109/ICRA46639.2022.9811650
ec_funded: 1
external_id:
arxiv:
- '2103.04909'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://doi.org/10.48550/arXiv.2103.04909
month: '07'
oa: 1
oa_version: Preprint
page: 7513-7520
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
publication: 2022 International Conference on Robotics and Automation
publication_identifier:
isbn:
- '9781728196817'
issn:
- 1050-4729
publication_status: published
publisher: IEEE
quality_controlled: '1'
scopus_import: '1'
status: public
title: Latent imagination facilitates zero-shot transfer in autonomous racing
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
year: '2022'
...
---
_id: '12147'
abstract:
- lang: eng
text: Continuous-time neural networks are a class of machine learning systems that
can tackle representation learning on spatiotemporal decision-making tasks. These
models are typically represented by continuous differential equations. However,
their expressive power when they are deployed on computers is bottlenecked by
numerical differential equation solvers. This limitation has notably slowed down
the scaling and understanding of numerous natural physical phenomena such as the
dynamics of nervous systems. Ideally, we would circumvent this bottleneck by solving
the given dynamical system in closed form. This is known to be intractable in
general. Here, we show that it is possible to closely approximate the interaction
between neurons and synapses—the building blocks of natural and artificial neural
networks—constructed by liquid time-constant networks efficiently in closed form.
To this end, we compute a tightly bounded approximation of the solution of an
integral appearing in liquid time-constant dynamics that has had no known closed-form
solution so far. This closed-form solution impacts the design of continuous-time
and continuous-depth neural models. For instance, since time appears explicitly
in closed form, the formulation relaxes the need for complex numerical solvers.
Consequently, we obtain models that are between one and five orders of magnitude
faster in training and inference compared with differential equation-based counterparts.
More importantly, in contrast to ordinary differential equation-based continuous
networks, closed-form networks can scale remarkably well compared with other deep
learning instances. Lastly, as these models are derived from liquid networks,
they show good performance in time-series modelling compared with advanced recurrent
neural network models.
acknowledgement: This research was supported in part by the AI2050 program at Schmidt
Futures (grant G-22-63172), the Boeing Company, and the United States Air Force
Research Laboratory and the United States Air Force Artificial Intelligence Accelerator
and was accomplished under cooperative agreement number FA8750-19-2-1000. The views
and conclusions contained in this document are those of the authors and should not
be interpreted as representing the official policies, either expressed or implied,
of the United States Air Force or the U.S. Government. The U.S. Government is authorized
to reproduce and distribute reprints for Government purposes, notwithstanding any
copyright notation herein. This work was further supported by The Boeing Company
and Office of Naval Research grant N00014-18-1-2830. M.T. is supported by the Poul
Due Jensen Foundation, grant 883901. M.L. was supported in part by the Austrian
Science Fund under grant Z211-N23 (Wittgenstein Award). A.A. was supported by the
National Science Foundation Graduate Research Fellowship Program. We thank T.-H.
Wang, P. Kao, M. Chahine, W. Xiao, X. Li, L. Yin and Y. Ben for useful suggestions
and for testing of CfC models to confirm the results across other domains.
article_processing_charge: No
article_type: original
arxiv: 1
author:
- first_name: Ramin
full_name: Hasani, Ramin
last_name: Hasani
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Alexander
full_name: Amini, Alexander
last_name: Amini
- first_name: Lucas
full_name: Liebenwein, Lucas
last_name: Liebenwein
- first_name: Aaron
full_name: Ray, Aaron
last_name: Ray
- first_name: Max
full_name: Tschaikowski, Max
last_name: Tschaikowski
- first_name: Gerald
full_name: Teschl, Gerald
last_name: Teschl
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
citation:
ama: Hasani R, Lechner M, Amini A, et al. Closed-form continuous-time neural networks.
Nature Machine Intelligence. 2022;4(11):992-1003. doi:10.1038/s42256-022-00556-7
apa: Hasani, R., Lechner, M., Amini, A., Liebenwein, L., Ray, A., Tschaikowski,
M., … Rus, D. (2022). Closed-form continuous-time neural networks. Nature Machine
Intelligence. Springer Nature. https://doi.org/10.1038/s42256-022-00556-7
chicago: Hasani, Ramin, Mathias Lechner, Alexander Amini, Lucas Liebenwein, Aaron
Ray, Max Tschaikowski, Gerald Teschl, and Daniela Rus. “Closed-Form Continuous-Time
Neural Networks.” Nature Machine Intelligence. Springer Nature, 2022. https://doi.org/10.1038/s42256-022-00556-7.
ieee: R. Hasani et al., “Closed-form continuous-time neural networks,” Nature
Machine Intelligence, vol. 4, no. 11. Springer Nature, pp. 992–1003, 2022.
ista: Hasani R, Lechner M, Amini A, Liebenwein L, Ray A, Tschaikowski M, Teschl
G, Rus D. 2022. Closed-form continuous-time neural networks. Nature Machine Intelligence.
4(11), 992–1003.
mla: Hasani, Ramin, et al. “Closed-Form Continuous-Time Neural Networks.” Nature
Machine Intelligence, vol. 4, no. 11, Springer Nature, 2022, pp. 992–1003,
doi:10.1038/s42256-022-00556-7.
short: R. Hasani, M. Lechner, A. Amini, L. Liebenwein, A. Ray, M. Tschaikowski,
G. Teschl, D. Rus, Nature Machine Intelligence 4 (2022) 992–1003.
date_created: 2023-01-12T12:07:21Z
date_published: 2022-11-15T00:00:00Z
date_updated: 2023-08-04T09:00:10Z
day: '15'
ddc:
- '000'
department:
- _id: ToHe
doi: 10.1038/s42256-022-00556-7
external_id:
arxiv:
- '2106.13898'
isi:
- '000884215600003'
file:
- access_level: open_access
checksum: b4789122ce04bfb4ac042390f59aaa8b
content_type: application/pdf
creator: dernst
date_created: 2023-01-24T09:49:44Z
date_updated: 2023-01-24T09:49:44Z
file_id: '12355'
file_name: 2022_NatureMachineIntelligence_Hasani.pdf
file_size: 3259553
relation: main_file
success: 1
file_date_updated: 2023-01-24T09:49:44Z
has_accepted_license: '1'
intvolume: ' 4'
isi: 1
issue: '11'
keyword:
- Artificial Intelligence
- Computer Networks and Communications
- Computer Vision and Pattern Recognition
- Human-Computer Interaction
- Software
language:
- iso: eng
month: '11'
oa: 1
oa_version: Published Version
page: 992-1003
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
publication: Nature Machine Intelligence
publication_identifier:
issn:
- 2522-5839
publication_status: published
publisher: Springer Nature
quality_controlled: '1'
related_material:
link:
- relation: erratum
url: https://doi.org/10.1038/s42256-022-00597-y
scopus_import: '1'
status: public
title: Closed-form continuous-time neural networks
tmp:
image: /images/cc_by.png
legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode
name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)
short: CC BY (4.0)
type: journal_article
user_id: 4359f0d1-fa6c-11eb-b949-802e58b17ae8
volume: 4
year: '2022'
...
---
_id: '12510'
abstract:
- lang: eng
text: "We introduce a new statistical verification algorithm that formally quantifies
the behavioral robustness of any time-continuous process formulated as a continuous-depth
model. Our algorithm solves a set of global optimization (Go) problems over a
given time horizon to construct a tight enclosure (Tube) of the set of all process
executions starting from a ball of initial states. We call our algorithm GoTube.
Through its construction, GoTube ensures that the bounding tube is conservative
up to a desired probability and up to a desired tightness.\r\n GoTube is implemented
in JAX and optimized to scale to complex continuous-depth neural network models.
Compared to advanced reachability analysis tools for time-continuous neural networks,
GoTube does not accumulate overapproximation errors between time steps and avoids
the infamous wrapping effect inherent in symbolic techniques. We show that GoTube
substantially outperforms state-of-the-art verification tools in terms of the
size of the initial ball, speed, time-horizon, task completion, and scalability
on a large set of experiments.\r\n GoTube is stable and sets the state-of-the-art
in terms of its ability to scale to time horizons well beyond what has been previously
possible."
acknowledgement: SG is funded by the Austrian Science Fund (FWF) project number W1255-N23.
ML and TH are supported in part by FWF under grant Z211-N23 (Wittgenstein Award)
and the ERC-2020-AdG 101020093. SS is supported by NSF awards DCL-2040599, CCF-1918225,
and CPS-1446832. RH and DR are partially supported by Boeing. RG is partially supported
by Horizon-2020 ECSEL Project grant No. 783163 (iDev40).
article_processing_charge: No
article_type: original
arxiv: 1
author:
- first_name: Sophie A.
full_name: Gruenbacher, Sophie A.
last_name: Gruenbacher
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Ramin
full_name: Hasani, Ramin
last_name: Hasani
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
- first_name: Scott A.
full_name: Smolka, Scott A.
last_name: Smolka
- first_name: Radu
full_name: Grosu, Radu
last_name: Grosu
citation:
ama: 'Gruenbacher SA, Lechner M, Hasani R, et al. GoTube: Scalable statistical verification
of continuous-depth models. Proceedings of the AAAI Conference on Artificial
Intelligence. 2022;36(6):6755-6764. doi:10.1609/aaai.v36i6.20631'
apa: 'Gruenbacher, S. A., Lechner, M., Hasani, R., Rus, D., Henzinger, T. A., Smolka,
S. A., & Grosu, R. (2022). GoTube: Scalable statistical verification of continuous-depth
models. Proceedings of the AAAI Conference on Artificial Intelligence.
Association for the Advancement of Artificial Intelligence. https://doi.org/10.1609/aaai.v36i6.20631'
chicago: 'Gruenbacher, Sophie A., Mathias Lechner, Ramin Hasani, Daniela Rus, Thomas
A Henzinger, Scott A. Smolka, and Radu Grosu. “GoTube: Scalable Statistical Verification
of Continuous-Depth Models.” Proceedings of the AAAI Conference on Artificial
Intelligence. Association for the Advancement of Artificial Intelligence,
2022. https://doi.org/10.1609/aaai.v36i6.20631.'
ieee: 'S. A. Gruenbacher et al., “GoTube: Scalable statistical verification
of continuous-depth models,” Proceedings of the AAAI Conference on Artificial
Intelligence, vol. 36, no. 6. Association for the Advancement of Artificial
Intelligence, pp. 6755–6764, 2022.'
ista: 'Gruenbacher SA, Lechner M, Hasani R, Rus D, Henzinger TA, Smolka SA, Grosu
R. 2022. GoTube: Scalable statistical verification of continuous-depth models.
Proceedings of the AAAI Conference on Artificial Intelligence. 36(6), 6755–6764.'
mla: 'Gruenbacher, Sophie A., et al. “GoTube: Scalable Statistical Verification
of Continuous-Depth Models.” Proceedings of the AAAI Conference on Artificial
Intelligence, vol. 36, no. 6, Association for the Advancement of Artificial
Intelligence, 2022, pp. 6755–64, doi:10.1609/aaai.v36i6.20631.'
short: S.A. Gruenbacher, M. Lechner, R. Hasani, D. Rus, T.A. Henzinger, S.A. Smolka,
R. Grosu, Proceedings of the AAAI Conference on Artificial Intelligence 36 (2022)
6755–6764.
date_created: 2023-02-05T17:27:42Z
date_published: 2022-06-28T00:00:00Z
date_updated: 2023-09-26T10:46:59Z
day: '28'
department:
- _id: ToHe
doi: 10.1609/aaai.v36i6.20631
ec_funded: 1
external_id:
arxiv:
- '2107.08467'
intvolume: ' 36'
issue: '6'
keyword:
- General Medicine
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://arxiv.org/abs/2107.08467
month: '06'
oa: 1
oa_version: Preprint
page: 6755-6764
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
publication: Proceedings of the AAAI Conference on Artificial Intelligence
publication_identifier:
eissn:
- 2374-3468
isbn:
- '978577358350'
issn:
- 2159-5399
publication_status: published
publisher: Association for the Advancement of Artificial Intelligence
quality_controlled: '1'
scopus_import: '1'
status: public
title: 'GoTube: Scalable statistical verification of continuous-depth models'
type: journal_article
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 36
year: '2022'
...
---
_id: '12511'
abstract:
- lang: eng
text: "We consider the problem of formally verifying almost-sure (a.s.) asymptotic
stability in discrete-time nonlinear stochastic control systems. While verifying
stability in deterministic control systems is extensively studied in the literature,
verifying stability in stochastic control systems is an open problem. The few
existing works on this topic either consider only specialized forms of stochasticity
or make restrictive assumptions on the system, rendering them inapplicable to
learning algorithms with neural network policies. \r\n In this work, we present
an approach for general nonlinear stochastic control problems with two novel aspects:
(a) instead of classical stochastic extensions of Lyapunov functions, we use ranking
supermartingales (RSMs) to certify a.s. asymptotic stability, and (b) we present
a method for learning neural network RSMs. \r\n We prove that our approach guarantees
a.s. asymptotic stability of the system and\r\n provides the first method to obtain
bounds on the stabilization time, which stochastic Lyapunov functions do not.\r\n
Finally, we validate our approach experimentally on a set of nonlinear stochastic
reinforcement learning environments with neural network policies."
acknowledgement: "This work was supported in part by the ERC-2020-AdG 101020093, ERC
CoG 863818 (FoRM-SMArt) and the European Union’s Horizon 2020 research and innovation
programme\r\nunder the Marie Skłodowska-Curie Grant Agreement No. 665385."
article_processing_charge: No
article_type: original
arxiv: 1
author:
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
citation:
ama: Lechner M, Zikelic D, Chatterjee K, Henzinger TA. Stability verification in
stochastic control systems via neural network supermartingales. Proceedings
of the AAAI Conference on Artificial Intelligence. 2022;36(7):7326-7336. doi:10.1609/aaai.v36i7.20695
apa: Lechner, M., Zikelic, D., Chatterjee, K., & Henzinger, T. A. (2022). Stability
verification in stochastic control systems via neural network supermartingales.
Proceedings of the AAAI Conference on Artificial Intelligence. Association
for the Advancement of Artificial Intelligence. https://doi.org/10.1609/aaai.v36i7.20695
chicago: Lechner, Mathias, Dorde Zikelic, Krishnendu Chatterjee, and Thomas A Henzinger.
“Stability Verification in Stochastic Control Systems via Neural Network Supermartingales.”
Proceedings of the AAAI Conference on Artificial Intelligence. Association
for the Advancement of Artificial Intelligence, 2022. https://doi.org/10.1609/aaai.v36i7.20695.
ieee: M. Lechner, D. Zikelic, K. Chatterjee, and T. A. Henzinger, “Stability verification
in stochastic control systems via neural network supermartingales,” Proceedings
of the AAAI Conference on Artificial Intelligence, vol. 36, no. 7. Association
for the Advancement of Artificial Intelligence, pp. 7326–7336, 2022.
ista: Lechner M, Zikelic D, Chatterjee K, Henzinger TA. 2022. Stability verification
in stochastic control systems via neural network supermartingales. Proceedings
of the AAAI Conference on Artificial Intelligence. 36(7), 7326–7336.
mla: Lechner, Mathias, et al. “Stability Verification in Stochastic Control Systems
via Neural Network Supermartingales.” Proceedings of the AAAI Conference on
Artificial Intelligence, vol. 36, no. 7, Association for the Advancement of
Artificial Intelligence, 2022, pp. 7326–36, doi:10.1609/aaai.v36i7.20695.
short: M. Lechner, D. Zikelic, K. Chatterjee, T.A. Henzinger, Proceedings of the
AAAI Conference on Artificial Intelligence 36 (2022) 7326–7336.
date_created: 2023-02-05T17:29:50Z
date_published: 2022-06-28T00:00:00Z
date_updated: 2025-07-14T09:09:58Z
day: '28'
department:
- _id: ToHe
- _id: KrCh
doi: 10.1609/aaai.v36i7.20695
ec_funded: 1
external_id:
arxiv:
- '2112.09495'
intvolume: ' 36'
issue: '7'
keyword:
- General Medicine
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://arxiv.org/abs/2112.09495
month: '06'
oa: 1
oa_version: Preprint
page: 7326-7336
project:
- _id: 62781420-2b32-11ec-9570-8d9b63373d4d
call_identifier: H2020
grant_number: '101020093'
name: Vigilant Algorithmic Monitoring of Software
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
publication: Proceedings of the AAAI Conference on Artificial Intelligence
publication_identifier:
eissn:
- 2374-3468
isbn:
- '9781577358350'
issn:
- 2159-5399
publication_status: published
publisher: Association for the Advancement of Artificial Intelligence
quality_controlled: '1'
related_material:
record:
- id: '14539'
relation: dissertation_contains
status: public
scopus_import: '1'
status: public
title: Stability verification in stochastic control systems via neural network supermartingales
type: journal_article
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 36
year: '2022'
...
---
_id: '10665'
abstract:
- lang: eng
text: "Formal verification of neural networks is an active topic of research, and
recent advances have significantly increased the size of the networks that verification
tools can handle. However, most methods are designed for verification of an idealized
model of the actual network which works over real arithmetic and ignores rounding
imprecisions. This idealization is in stark contrast to network quantization,
which is a technique that trades numerical precision for computational efficiency
and is, therefore, often applied in practice. Neglecting rounding errors of such
low-bit quantized neural networks has been shown to lead to wrong conclusions
about the network’s correctness. Thus, the desired approach for verifying quantized
neural networks would be one that takes these rounding errors\r\ninto account.
In this paper, we show that verifying the bitexact implementation of quantized
neural networks with bitvector specifications is PSPACE-hard, even though verifying
idealized real-valued networks and satisfiability of bit-vector specifications
alone are each in NP. Furthermore, we explore several practical heuristics toward
closing the complexity gap between idealized and bit-exact verification. In particular,
we propose three techniques for making SMT-based verification of quantized neural
networks more scalable. Our experiments demonstrate that our proposed methods
allow a speedup of up to three orders of magnitude over existing approaches."
acknowledgement: "This research was supported in part by the Austrian Science Fund
(FWF) under grant Z211-N23 (Wittgenstein\r\nAward), ERC CoG 863818 (FoRM-SMArt),
and the European Union’s Horizon 2020 research and innovation programme under the
Marie Skłodowska-Curie Grant Agreement No. 665385.\r\n"
alternative_title:
- Technical Tracks
article_processing_charge: No
arxiv: 1
author:
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Dorde
full_name: Zikelic, Dorde
id: 294AA7A6-F248-11E8-B48F-1D18A9856A87
last_name: Zikelic
orcid: 0000-0002-4681-1699
citation:
ama: 'Henzinger TA, Lechner M, Zikelic D. Scalable verification of quantized neural
networks. In: Proceedings of the AAAI Conference on Artificial Intelligence.
Vol 35. AAAI Press; 2021:3787-3795.'
apa: 'Henzinger, T. A., Lechner, M., & Zikelic, D. (2021). Scalable verification
of quantized neural networks. In Proceedings of the AAAI Conference on Artificial
Intelligence (Vol. 35, pp. 3787–3795). Virtual: AAAI Press.'
chicago: Henzinger, Thomas A, Mathias Lechner, and Dorde Zikelic. “Scalable Verification
of Quantized Neural Networks.” In Proceedings of the AAAI Conference on Artificial
Intelligence, 35:3787–95. AAAI Press, 2021.
ieee: T. A. Henzinger, M. Lechner, and D. Zikelic, “Scalable verification of quantized
neural networks,” in Proceedings of the AAAI Conference on Artificial Intelligence,
Virtual, 2021, vol. 35, no. 5A, pp. 3787–3795.
ista: 'Henzinger TA, Lechner M, Zikelic D. 2021. Scalable verification of quantized
neural networks. Proceedings of the AAAI Conference on Artificial Intelligence.
AAAI: Association for the Advancement of Artificial Intelligence, Technical Tracks,
vol. 35, 3787–3795.'
mla: Henzinger, Thomas A., et al. “Scalable Verification of Quantized Neural Networks.”
Proceedings of the AAAI Conference on Artificial Intelligence, vol. 35,
no. 5A, AAAI Press, 2021, pp. 3787–95.
short: T.A. Henzinger, M. Lechner, D. Zikelic, in:, Proceedings of the AAAI Conference
on Artificial Intelligence, AAAI Press, 2021, pp. 3787–3795.
conference:
end_date: 2021-02-09
location: Virtual
name: 'AAAI: Association for the Advancement of Artificial Intelligence'
start_date: 2021-02-02
date_created: 2022-01-25T15:15:02Z
date_published: 2021-05-28T00:00:00Z
date_updated: 2025-07-14T09:10:11Z
day: '28'
ddc:
- '000'
department:
- _id: GradSch
- _id: ToHe
ec_funded: 1
external_id:
arxiv:
- '2012.08185'
file:
- access_level: open_access
checksum: 2bc8155b2526a70fba5b7301bc89dbd1
content_type: application/pdf
creator: mlechner
date_created: 2022-01-26T07:41:16Z
date_updated: 2022-01-26T07:41:16Z
file_id: '10684'
file_name: 16496-Article Text-19990-1-2-20210518 (1).pdf
file_size: 137235
relation: main_file
success: 1
file_date_updated: 2022-01-26T07:41:16Z
has_accepted_license: '1'
intvolume: ' 35'
issue: 5A
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://ojs.aaai.org/index.php/AAAI/article/view/16496
month: '05'
oa: 1
oa_version: Published Version
page: 3787-3795
project:
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
publication: Proceedings of the AAAI Conference on Artificial Intelligence
publication_identifier:
eissn:
- 2374-3468
isbn:
- 978-1-57735-866-4
issn:
- 2159-5399
publication_status: published
publisher: AAAI Press
quality_controlled: '1'
related_material:
record:
- id: '11362'
relation: dissertation_contains
status: public
scopus_import: '1'
status: public
title: Scalable verification of quantized neural networks
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 35
year: '2021'
...
---
_id: '10666'
abstract:
- lang: eng
text: Adversarial training is an effective method to train deep learning models
that are resilient to norm-bounded perturbations, with the cost of nominal performance
drop. While adversarial training appears to enhance the robustness and safety
of a deep model deployed in open-world decision-critical applications, counterintuitively,
it induces undesired behaviors in robot learning settings. In this paper, we show
theoretically and experimentally that neural controllers obtained via adversarial
training are subjected to three types of defects, namely transient, systematic,
and conditional errors. We first generalize adversarial training to a safety-domain
optimization scheme allowing for more generic specifications. We then prove that
such a learning process tends to cause certain error profiles. We support our
theoretical results by a thorough experimental safety analysis in a robot-learning
task. Our results suggest that adversarial training is not yet ready for robot
learning.
acknowledgement: M.L. and T.A.H. are supported in part by the Austrian Science Fund
(FWF) under grant Z211-N23 (Wittgenstein Award). R.H. and D.R. are supported by
Boeing and R.G. by Horizon-2020 ECSEL Project grant no. 783163 (iDev40).
article_processing_charge: No
arxiv: 1
author:
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Ramin
full_name: Hasani, Ramin
last_name: Hasani
- first_name: Radu
full_name: Grosu, Radu
last_name: Grosu
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
citation:
ama: 'Lechner M, Hasani R, Grosu R, Rus D, Henzinger TA. Adversarial training is
not ready for robot learning. In: 2021 IEEE International Conference on Robotics
and Automation. ICRA. ; 2021:4140-4147. doi:10.1109/ICRA48506.2021.9561036'
apa: Lechner, M., Hasani, R., Grosu, R., Rus, D., & Henzinger, T. A. (2021).
Adversarial training is not ready for robot learning. In 2021 IEEE International
Conference on Robotics and Automation (pp. 4140–4147). Xi’an, China. https://doi.org/10.1109/ICRA48506.2021.9561036
chicago: Lechner, Mathias, Ramin Hasani, Radu Grosu, Daniela Rus, and Thomas A Henzinger.
“Adversarial Training Is Not Ready for Robot Learning.” In 2021 IEEE International
Conference on Robotics and Automation, 4140–47. ICRA, 2021. https://doi.org/10.1109/ICRA48506.2021.9561036.
ieee: M. Lechner, R. Hasani, R. Grosu, D. Rus, and T. A. Henzinger, “Adversarial
training is not ready for robot learning,” in 2021 IEEE International Conference
on Robotics and Automation, Xi’an, China, 2021, pp. 4140–4147.
ista: 'Lechner M, Hasani R, Grosu R, Rus D, Henzinger TA. 2021. Adversarial training
is not ready for robot learning. 2021 IEEE International Conference on Robotics
and Automation. ICRA: International Conference on Robotics and AutomationICRA,
4140–4147.'
mla: Lechner, Mathias, et al. “Adversarial Training Is Not Ready for Robot Learning.”
2021 IEEE International Conference on Robotics and Automation, 2021, pp.
4140–47, doi:10.1109/ICRA48506.2021.9561036.
short: M. Lechner, R. Hasani, R. Grosu, D. Rus, T.A. Henzinger, in:, 2021 IEEE International
Conference on Robotics and Automation, 2021, pp. 4140–4147.
conference:
end_date: 2021-06-05
location: Xi'an, China
name: 'ICRA: International Conference on Robotics and Automation'
start_date: 2021-05-30
date_created: 2022-01-25T15:44:54Z
date_published: 2021-01-01T00:00:00Z
date_updated: 2023-08-17T06:58:38Z
ddc:
- '000'
department:
- _id: GradSch
- _id: ToHe
doi: 10.1109/ICRA48506.2021.9561036
external_id:
arxiv:
- '2103.08187'
isi:
- '000765738803040'
has_accepted_license: '1'
isi: 1
language:
- iso: eng
license: https://creativecommons.org/licenses/by-nc-nd/3.0/
main_file_link:
- open_access: '1'
url: https://arxiv.org/abs/2103.08187
oa: 1
oa_version: None
page: 4140-4147
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
publication: 2021 IEEE International Conference on Robotics and Automation
publication_identifier:
eisbn:
- 978-1-7281-9077-8
eissn:
- 2577-087X
isbn:
- 978-1-7281-9078-5
issn:
- 1050-4729
publication_status: published
quality_controlled: '1'
related_material:
record:
- id: '11362'
relation: dissertation_contains
status: public
series_title: ICRA
status: public
title: Adversarial training is not ready for robot learning
tmp:
image: /images/cc_by_nc_nd.png
legal_code_url: https://creativecommons.org/licenses/by-nc-nd/3.0/legalcode
name: Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported (CC BY-NC-ND
3.0)
short: CC BY-NC-ND (3.0)
type: conference
user_id: 4359f0d1-fa6c-11eb-b949-802e58b17ae8
year: '2021'
...
---
_id: '10667'
abstract:
- lang: eng
text: Bayesian neural networks (BNNs) place distributions over the weights of a
neural network to model uncertainty in the data and the network's prediction.
We consider the problem of verifying safety when running a Bayesian neural network
policy in a feedback loop with infinite time horizon systems. Compared to the
existing sampling-based approaches, which are inapplicable to the infinite time
horizon setting, we train a separate deterministic neural network that serves
as an infinite time horizon safety certificate. In particular, we show that the
certificate network guarantees the safety of the system over a subset of the BNN
weight posterior's support. Our method first computes a safe weight set and then
alters the BNN's weight posterior to reject samples outside this set. Moreover,
we show how to extend our approach to a safe-exploration reinforcement learning
setting, in order to avoid unsafe trajectories during the training of the policy.
We evaluate our approach on a series of reinforcement learning benchmarks, including
non-Lyapunovian safety specifications.
acknowledgement: This research was supported in part by the Austrian Science Fund
(FWF) under grant Z211-N23 (Wittgenstein Award), ERC CoG 863818 (FoRM-SMArt), and
the European Union’s Horizon 2020 research and innovation programme under the Marie
Skłodowska-Curie Grant Agreement No. 665385.
alternative_title:
- ' Advances in Neural Information Processing Systems'
article_processing_charge: No
arxiv: 1
author:
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Ðorđe
full_name: Žikelić, Ðorđe
last_name: Žikelić
- first_name: Krishnendu
full_name: Chatterjee, Krishnendu
id: 2E5DCA20-F248-11E8-B48F-1D18A9856A87
last_name: Chatterjee
orcid: 0000-0002-4561-241X
- first_name: Thomas A
full_name: Henzinger, Thomas A
id: 40876CD8-F248-11E8-B48F-1D18A9856A87
last_name: Henzinger
orcid: 0000-0002-2985-7724
citation:
ama: 'Lechner M, Žikelić Ð, Chatterjee K, Henzinger TA. Infinite time horizon safety
of Bayesian neural networks. In: 35th Conference on Neural Information Processing
Systems. ; 2021. doi:10.48550/arXiv.2111.03165'
apa: Lechner, M., Žikelić, Ð., Chatterjee, K., & Henzinger, T. A. (2021). Infinite
time horizon safety of Bayesian neural networks. In 35th Conference on Neural
Information Processing Systems. Virtual. https://doi.org/10.48550/arXiv.2111.03165
chicago: Lechner, Mathias, Ðorđe Žikelić, Krishnendu Chatterjee, and Thomas A Henzinger.
“Infinite Time Horizon Safety of Bayesian Neural Networks.” In 35th Conference
on Neural Information Processing Systems, 2021. https://doi.org/10.48550/arXiv.2111.03165.
ieee: M. Lechner, Ð. Žikelić, K. Chatterjee, and T. A. Henzinger, “Infinite time
horizon safety of Bayesian neural networks,” in 35th Conference on Neural Information
Processing Systems, Virtual, 2021.
ista: 'Lechner M, Žikelić Ð, Chatterjee K, Henzinger TA. 2021. Infinite time horizon
safety of Bayesian neural networks. 35th Conference on Neural Information Processing
Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information
Processing Systems, .'
mla: Lechner, Mathias, et al. “Infinite Time Horizon Safety of Bayesian Neural Networks.”
35th Conference on Neural Information Processing Systems, 2021, doi:10.48550/arXiv.2111.03165.
short: M. Lechner, Ð. Žikelić, K. Chatterjee, T.A. Henzinger, in:, 35th Conference
on Neural Information Processing Systems, 2021.
conference:
end_date: 2021-12-10
location: Virtual
name: 'NeurIPS: Neural Information Processing Systems'
start_date: 2021-12-06
date_created: 2022-01-25T15:45:58Z
date_published: 2021-12-01T00:00:00Z
date_updated: 2025-07-14T09:10:12Z
day: '01'
ddc:
- '000'
department:
- _id: GradSch
- _id: ToHe
- _id: KrCh
doi: 10.48550/arXiv.2111.03165
ec_funded: 1
external_id:
arxiv:
- '2111.03165'
file:
- access_level: open_access
checksum: 0fc0f852525c10dda9cc9ffea07fb4e4
content_type: application/pdf
creator: mlechner
date_created: 2022-01-26T07:39:59Z
date_updated: 2022-01-26T07:39:59Z
file_id: '10682'
file_name: infinite_time_horizon_safety_o.pdf
file_size: 452492
relation: main_file
success: 1
file_date_updated: 2022-01-26T07:39:59Z
has_accepted_license: '1'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://proceedings.neurips.cc/paper/2021/hash/544defa9fddff50c53b71c43e0da72be-Abstract.html
month: '12'
oa: 1
oa_version: Published Version
project:
- _id: 2564DBCA-B435-11E9-9278-68D0E5697425
call_identifier: H2020
grant_number: '665385'
name: International IST Doctoral Program
- _id: 0599E47C-7A3F-11EA-A408-12923DDC885E
call_identifier: H2020
grant_number: '863818'
name: 'Formal Methods for Stochastic Models: Algorithms and Applications'
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
publication: 35th Conference on Neural Information Processing Systems
publication_status: published
quality_controlled: '1'
related_material:
record:
- id: '11362'
relation: dissertation_contains
status: public
status: public
title: Infinite time horizon safety of Bayesian neural networks
tmp:
image: /images/cc_by_nc_nd.png
legal_code_url: https://creativecommons.org/licenses/by-nc-nd/3.0/legalcode
name: Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported (CC BY-NC-ND
3.0)
short: CC BY-NC-ND (3.0)
type: conference
user_id: 2EBD1598-F248-11E8-B48F-1D18A9856A87
year: '2021'
...
---
_id: '10668'
abstract:
- lang: eng
text: 'Robustness to variations in lighting conditions is a key objective for any
deep vision system. To this end, our paper extends the receptive field of convolutional
neural networks with two residual components, ubiquitous in the visual processing
system of vertebrates: On-center and off-center pathways, with an excitatory center
and inhibitory surround; OOCS for short. The On-center pathway is excited by the
presence of a light stimulus in its center, but not in its surround, whereas the
Off-center pathway is excited by the absence of a light stimulus in its center,
but not in its surround. We design OOCS pathways via a difference of Gaussians,
with their variance computed analytically from the size of the receptive fields.
OOCS pathways complement each other in their response to light stimuli, ensuring
this way a strong edge-detection capability, and as a result an accurate and robust
inference under challenging lighting conditions. We provide extensive empirical
evidence showing that networks supplied with OOCS pathways gain accuracy and illumination-robustness
from the novel edge representation, compared to other baselines.'
acknowledgement: Z.B. is supported by the Doctoral College Resilient Embedded Systems,
which is run jointly by the TU Wien’s Faculty of Informatics and the UAS Technikum
Wien. R.G. is partially supported by the Horizon 2020 Era-Permed project Persorad,
and ECSEL Project grant no. 783163 (iDev40). R.H and D.R were partially supported
by Boeing and MIT. M.L. is supported in part by the Austrian Science Fund (FWF)
under grant Z211-N23 (Wittgenstein Award).
alternative_title:
- PMLR
article_processing_charge: No
author:
- first_name: Zahra
full_name: Babaiee, Zahra
last_name: Babaiee
- first_name: Ramin
full_name: Hasani, Ramin
last_name: Hasani
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
- first_name: Radu
full_name: Grosu, Radu
last_name: Grosu
citation:
ama: 'Babaiee Z, Hasani R, Lechner M, Rus D, Grosu R. On-off center-surround receptive
fields for accurate and robust image classification. In: Proceedings of the
38th International Conference on Machine Learning. Vol 139. ML Research Press;
2021:478-489.'
apa: 'Babaiee, Z., Hasani, R., Lechner, M., Rus, D., & Grosu, R. (2021). On-off
center-surround receptive fields for accurate and robust image classification.
In Proceedings of the 38th International Conference on Machine Learning
(Vol. 139, pp. 478–489). Virtual: ML Research Press.'
chicago: Babaiee, Zahra, Ramin Hasani, Mathias Lechner, Daniela Rus, and Radu Grosu.
“On-off Center-Surround Receptive Fields for Accurate and Robust Image Classification.”
In Proceedings of the 38th International Conference on Machine Learning,
139:478–89. ML Research Press, 2021.
ieee: Z. Babaiee, R. Hasani, M. Lechner, D. Rus, and R. Grosu, “On-off center-surround
receptive fields for accurate and robust image classification,” in Proceedings
of the 38th International Conference on Machine Learning, Virtual, 2021, vol.
139, pp. 478–489.
ista: 'Babaiee Z, Hasani R, Lechner M, Rus D, Grosu R. 2021. On-off center-surround
receptive fields for accurate and robust image classification. Proceedings of
the 38th International Conference on Machine Learning. ML: Machine Learning, PMLR,
vol. 139, 478–489.'
mla: Babaiee, Zahra, et al. “On-off Center-Surround Receptive Fields for Accurate
and Robust Image Classification.” Proceedings of the 38th International Conference
on Machine Learning, vol. 139, ML Research Press, 2021, pp. 478–89.
short: Z. Babaiee, R. Hasani, M. Lechner, D. Rus, R. Grosu, in:, Proceedings of
the 38th International Conference on Machine Learning, ML Research Press, 2021,
pp. 478–489.
conference:
end_date: 2021-07-24
location: Virtual
name: 'ML: Machine Learning'
start_date: 2021-07-18
date_created: 2022-01-25T15:46:33Z
date_published: 2021-07-01T00:00:00Z
date_updated: 2022-05-04T15:02:27Z
day: '01'
ddc:
- '000'
department:
- _id: GradSch
- _id: ToHe
file:
- access_level: open_access
checksum: d30eae62561bb517d9f978437d7677db
content_type: application/pdf
creator: mlechner
date_created: 2022-01-26T07:38:32Z
date_updated: 2022-01-26T07:38:32Z
file_id: '10681'
file_name: babaiee21a.pdf
file_size: 4246561
relation: main_file
success: 1
file_date_updated: 2022-01-26T07:38:32Z
has_accepted_license: '1'
intvolume: ' 139'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://proceedings.mlr.press/v139/babaiee21a
month: '07'
oa: 1
oa_version: Published Version
page: 478-489
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
publication: Proceedings of the 38th International Conference on Machine Learning
publication_identifier:
issn:
- 2640-3498
publication_status: published
publisher: ML Research Press
quality_controlled: '1'
status: public
title: On-off center-surround receptive fields for accurate and robust image classification
tmp:
image: /images/cc_by_nc_nd.png
legal_code_url: https://creativecommons.org/licenses/by-nc-nd/3.0/legalcode
name: Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported (CC BY-NC-ND
3.0)
short: CC BY-NC-ND (3.0)
type: conference
user_id: 2EBD1598-F248-11E8-B48F-1D18A9856A87
volume: 139
year: '2021'
...
---
_id: '10669'
abstract:
- lang: eng
text: "We show that Neural ODEs, an emerging class of timecontinuous neural networks,
can be verified by solving a set of global-optimization problems. For this purpose,
we introduce Stochastic Lagrangian Reachability (SLR), an\r\nabstraction-based
technique for constructing a tight Reachtube (an over-approximation of the set
of reachable states\r\nover a given time-horizon), and provide stochastic guarantees
in the form of confidence intervals for the Reachtube bounds. SLR inherently avoids
the infamous wrapping effect (accumulation of over-approximation errors) by performing
local optimization steps to expand safe regions instead of repeatedly forward-propagating
them as is done by deterministic reachability methods. To enable fast local optimizations,
we introduce a novel forward-mode adjoint sensitivity method to compute gradients
without the need for backpropagation. Finally, we establish asymptotic and non-asymptotic
convergence rates for SLR."
acknowledgement: "The authors would like to thank the reviewers for their insightful
comments. RH and RG were partially supported by\r\nHorizon-2020 ECSEL Project grant
No. 783163 (iDev40). RH was partially supported by Boeing. ML was supported\r\nin
part by the Austrian Science Fund (FWF) under grant Z211-N23 (Wittgenstein Award).
SG was funded by FWF\r\nproject W1255-N23. JC was partially supported by NAWA Polish
Returns grant PPN/PPO/2018/1/00029. SS was supported by NSF awards DCL-2040599,
CCF-1918225, and CPS-1446832.\r\n"
alternative_title:
- Technical Tracks
article_processing_charge: No
arxiv: 1
author:
- first_name: Sophie
full_name: Grunbacher, Sophie
last_name: Grunbacher
- first_name: Ramin
full_name: Hasani, Ramin
last_name: Hasani
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Jacek
full_name: Cyranka, Jacek
last_name: Cyranka
- first_name: Scott A
full_name: Smolka, Scott A
last_name: Smolka
- first_name: Radu
full_name: Grosu, Radu
last_name: Grosu
citation:
ama: 'Grunbacher S, Hasani R, Lechner M, Cyranka J, Smolka SA, Grosu R. On the verification
of neural ODEs with stochastic guarantees. In: Proceedings of the AAAI Conference
on Artificial Intelligence. Vol 35. AAAI Press; 2021:11525-11535.'
apa: 'Grunbacher, S., Hasani, R., Lechner, M., Cyranka, J., Smolka, S. A., &
Grosu, R. (2021). On the verification of neural ODEs with stochastic guarantees.
In Proceedings of the AAAI Conference on Artificial Intelligence (Vol.
35, pp. 11525–11535). Virtual: AAAI Press.'
chicago: Grunbacher, Sophie, Ramin Hasani, Mathias Lechner, Jacek Cyranka, Scott
A Smolka, and Radu Grosu. “On the Verification of Neural ODEs with Stochastic
Guarantees.” In Proceedings of the AAAI Conference on Artificial Intelligence,
35:11525–35. AAAI Press, 2021.
ieee: S. Grunbacher, R. Hasani, M. Lechner, J. Cyranka, S. A. Smolka, and R. Grosu,
“On the verification of neural ODEs with stochastic guarantees,” in Proceedings
of the AAAI Conference on Artificial Intelligence, Virtual, 2021, vol. 35,
no. 13, pp. 11525–11535.
ista: 'Grunbacher S, Hasani R, Lechner M, Cyranka J, Smolka SA, Grosu R. 2021. On
the verification of neural ODEs with stochastic guarantees. Proceedings of the
AAAI Conference on Artificial Intelligence. AAAI: Association for the Advancement
of Artificial Intelligence, Technical Tracks, vol. 35, 11525–11535.'
mla: Grunbacher, Sophie, et al. “On the Verification of Neural ODEs with Stochastic
Guarantees.” Proceedings of the AAAI Conference on Artificial Intelligence,
vol. 35, no. 13, AAAI Press, 2021, pp. 11525–35.
short: S. Grunbacher, R. Hasani, M. Lechner, J. Cyranka, S.A. Smolka, R. Grosu,
in:, Proceedings of the AAAI Conference on Artificial Intelligence, AAAI Press,
2021, pp. 11525–11535.
conference:
end_date: 2021-02-09
location: Virtual
name: 'AAAI: Association for the Advancement of Artificial Intelligence'
start_date: 2021-02-02
date_created: 2022-01-25T15:47:20Z
date_published: 2021-05-28T00:00:00Z
date_updated: 2022-05-24T06:33:14Z
day: '28'
ddc:
- '000'
department:
- _id: GradSch
- _id: ToHe
external_id:
arxiv:
- '2012.08863'
file:
- access_level: open_access
checksum: 468d07041e282a1d46ffdae92f709630
content_type: application/pdf
creator: mlechner
date_created: 2022-01-26T07:38:08Z
date_updated: 2022-01-26T07:38:08Z
file_id: '10680'
file_name: 17372-Article Text-20866-1-2-20210518.pdf
file_size: 286906
relation: main_file
success: 1
file_date_updated: 2022-01-26T07:38:08Z
has_accepted_license: '1'
intvolume: ' 35'
issue: '13'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://ojs.aaai.org/index.php/AAAI/article/view/17372
month: '05'
oa: 1
oa_version: Published Version
page: 11525-11535
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
publication: Proceedings of the AAAI Conference on Artificial Intelligence
publication_identifier:
eissn:
- 2374-3468
isbn:
- 978-1-57735-866-4
issn:
- 2159-5399
publication_status: published
publisher: AAAI Press
quality_controlled: '1'
status: public
title: On the verification of neural ODEs with stochastic guarantees
type: conference
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 35
year: '2021'
...
---
_id: '10670'
abstract:
- lang: eng
text: "Imitation learning enables high-fidelity, vision-based learning of policies
within rich, photorealistic environments. However, such techniques often rely
on traditional discrete-time neural models and face difficulties in generalizing
to domain shifts by failing to account for the causal relationships between the
agent and the environment. In this paper, we propose a theoretical and experimental
framework for learning causal representations using continuous-time neural networks,
specifically over their discrete-time counterparts. We evaluate our method in
the context of visual-control learning of drones over a series of complex tasks,
ranging from short- and long-term navigation, to chasing static and dynamic objects
through photorealistic environments. Our results demonstrate that causal continuous-time\r\ndeep
models can perform robust navigation tasks, where advanced recurrent models fail.
These models learn complex causal control representations directly from raw visual
inputs and scale to solve a variety of tasks using imitation learning."
acknowledgement: "C.V., R.H. A.A. and D.R. are partially supported by Boeing and MIT.
A.A. is supported by the National Science Foundation (NSF) Graduate Research Fellowship
Program. M.L. is supported in part by the Austrian Science Fund (FWF) under grant
Z211-N23 (Wittgenstein Award). Research was sponsored by the United States Air Force
Research Laboratory and the United States Air Force Artificial Intelligence Accelerator
and was accomplished under Cooperative Agreement Number FA8750-19-2-1000. The views
and conclusions contained in this document are those of the authors\r\nand should
not be interpreted as representing the official policies, either expressed or implied,
of the United States Air Force or the U.S. Government. The U.S. Government is authorized
to reproduce and distribute reprints for Government purposes notwithstanding any
copyright notation herein.\r\n"
alternative_title:
- ' Advances in Neural Information Processing Systems'
article_processing_charge: No
arxiv: 1
author:
- first_name: Charles J
full_name: Vorbach, Charles J
last_name: Vorbach
- first_name: Ramin
full_name: Hasani, Ramin
last_name: Hasani
- first_name: Alexander
full_name: Amini, Alexander
last_name: Amini
- first_name: Mathias
full_name: Lechner, Mathias
id: 3DC22916-F248-11E8-B48F-1D18A9856A87
last_name: Lechner
- first_name: Daniela
full_name: Rus, Daniela
last_name: Rus
citation:
ama: 'Vorbach CJ, Hasani R, Amini A, Lechner M, Rus D. Causal navigation by continuous-time
neural networks. In: 35th Conference on Neural Information Processing Systems.
; 2021.'
apa: Vorbach, C. J., Hasani, R., Amini, A., Lechner, M., & Rus, D. (2021). Causal
navigation by continuous-time neural networks. In 35th Conference on Neural
Information Processing Systems. Virtual.
chicago: Vorbach, Charles J, Ramin Hasani, Alexander Amini, Mathias Lechner, and
Daniela Rus. “Causal Navigation by Continuous-Time Neural Networks.” In 35th
Conference on Neural Information Processing Systems, 2021.
ieee: C. J. Vorbach, R. Hasani, A. Amini, M. Lechner, and D. Rus, “Causal navigation
by continuous-time neural networks,” in 35th Conference on Neural Information
Processing Systems, Virtual, 2021.
ista: 'Vorbach CJ, Hasani R, Amini A, Lechner M, Rus D. 2021. Causal navigation
by continuous-time neural networks. 35th Conference on Neural Information Processing
Systems. NeurIPS: Neural Information Processing Systems, Advances in Neural Information
Processing Systems, .'
mla: Vorbach, Charles J., et al. “Causal Navigation by Continuous-Time Neural Networks.”
35th Conference on Neural Information Processing Systems, 2021.
short: C.J. Vorbach, R. Hasani, A. Amini, M. Lechner, D. Rus, in:, 35th Conference
on Neural Information Processing Systems, 2021.
conference:
end_date: 2021-12-10
location: Virtual
name: 'NeurIPS: Neural Information Processing Systems'
start_date: 2021-12-06
date_created: 2022-01-25T15:47:50Z
date_published: 2021-12-01T00:00:00Z
date_updated: 2022-01-26T14:33:31Z
day: '01'
ddc:
- '000'
department:
- _id: GradSch
- _id: ToHe
external_id:
arxiv:
- '2106.08314'
file:
- access_level: open_access
checksum: be81f0ade174a8c9b2d4fe09590b2021
content_type: application/pdf
creator: mlechner
date_created: 2022-01-26T07:37:24Z
date_updated: 2022-01-26T07:37:24Z
file_id: '10679'
file_name: NeurIPS-2021-causal-navigation-by-continuous-time-neural-networks-Paper.pdf
file_size: 6841228
relation: main_file
success: 1
file_date_updated: 2022-01-26T07:37:24Z
has_accepted_license: '1'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://proceedings.neurips.cc/paper/2021/hash/67ba02d73c54f0b83c05507b7fb7267f-Abstract.html
month: '12'
oa: 1
oa_version: Published Version
project:
- _id: 25F42A32-B435-11E9-9278-68D0E5697425
call_identifier: FWF
grant_number: Z211
name: The Wittgenstein Prize
publication: 35th Conference on Neural Information Processing Systems
publication_status: published
quality_controlled: '1'
status: public
title: Causal navigation by continuous-time neural networks
tmp:
image: /images/cc_by_nc_nd.png
legal_code_url: https://creativecommons.org/licenses/by-nc-nd/3.0/legalcode
name: Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported (CC BY-NC-ND
3.0)
short: CC BY-NC-ND (3.0)
type: conference
user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9
year: '2021'
...