@article{9208, abstract = {Bending-active structures are able to efficiently produce complex curved shapes from flat panels. The desired deformation of the panels derives from the proper selection of their elastic properties. Optimized panels, called FlexMaps, are designed such that, once they are bent and assembled, the resulting static equilibrium configuration matches a desired input 3D shape. The FlexMaps elastic properties are controlled by locally varying spiraling geometric mesostructures, which are optimized in size and shape to match specific bending requests, namely the global curvature of the target shape. The design pipeline starts from a quad mesh representing the input 3D shape, which defines the edge size and the total amount of spirals: every quad will embed one spiral. Then, an optimization algorithm tunes the geometry of the spirals by using a simplified pre-computed rod model. This rod model is derived from a non-linear regression algorithm which approximates the non-linear behavior of solid FEM spiral models subject to hundreds of load combinations. This innovative pipeline has been applied to the project of a lightweight plywood pavilion named FlexMaps Pavilion, which is a single-layer piecewise twisted arch that fits a bounding box of 3.90x3.96x3.25 meters. This case study serves to test the applicability of this methodology at the architectural scale. The structure is validated via FE analyses and the fabrication of the full scale prototype.}, author = {Laccone, Francesco and Malomo, Luigi and Perez Rodriguez, Jesus and Pietroni, Nico and Ponchio, Federico and Bickel, Bernd and Cignoni, Paolo}, issn = {25233971}, journal = {SN Applied Sciences}, number = {9}, publisher = {Springer Nature}, title = {{A bending-active twisted-arch plywood structure: Computational design and fabrication of the FlexMaps Pavilion}}, doi = {10.1007/s42452-020-03305-w}, volume = {2}, year = {2020}, } @inproceedings{9221, abstract = {Recent works have shown that gradient descent can find a global minimum for over-parameterized neural networks where the widths of all the hidden layers scale polynomially with N (N being the number of training samples). In this paper, we prove that, for deep networks, a single layer of width N following the input layer suffices to ensure a similar guarantee. In particular, all the remaining layers are allowed to have constant widths, and form a pyramidal topology. We show an application of our result to the widely used LeCun’s initialization and obtain an over-parameterization requirement for the single wide layer of order N2. }, author = {Nguyen, Quynh and Mondelli, Marco}, booktitle = {34th Conference on Neural Information Processing Systems}, location = {Vancouver, Canada}, pages = {11961–11972}, publisher = {Curran Associates}, title = {{Global convergence of deep networks with one wide layer followed by pyramidal topology}}, volume = {33}, year = {2020}, } @misc{9222, author = {Katsaros, Georgios}, publisher = {Institute of Science and Technology Austria}, title = {{Transport data for: Site‐controlled uniform Ge/Si Hut wires with electrically tunable spin–orbit coupling}}, doi = {10.15479/AT:ISTA:9222}, year = {2020}, } @article{9249, abstract = {Rhombic dodecahedron is a space filling polyhedron which represents the close packing of spheres in 3D space and the Voronoi structures of the face centered cubic (FCC) lattice. In this paper, we describe a new coordinate system where every 3-integer coordinates grid point corresponds to a rhombic dodecahedron centroid. In order to illustrate the interest of the new coordinate system, we propose the characterization of 3D digital plane with its topological features, such as the interrelation between the thickness of the digital plane and the separability constraint we aim to obtain. We also present the characterization of 3D digital lines and study it as the intersection of multiple digital planes. Characterization of 3D digital sphere with relevant topological features is proposed as well along with the 48-symmetry appearing in the new coordinate system.}, author = {Biswas, Ranita and Largeteau-Skapin, Gaëlle and Zrour, Rita and Andres, Eric}, issn = {2353-3390}, journal = {Mathematical Morphology - Theory and Applications}, number = {1}, pages = {143--158}, publisher = {De Gruyter}, title = {{Digital objects in rhombic dodecahedron grid}}, doi = {10.1515/mathm-2020-0106}, volume = {4}, year = {2020}, } @inproceedings{9299, abstract = {We call a multigraph non-homotopic if it can be drawn in the plane in such a way that no two edges connecting the same pair of vertices can be continuously transformed into each other without passing through a vertex, and no loop can be shrunk to its end-vertex in the same way. It is easy to see that a non-homotopic multigraph on n>1 vertices can have arbitrarily many edges. We prove that the number of crossings between the edges of a non-homotopic multigraph with n vertices and m>4n edges is larger than cm2n for some constant c>0 , and that this bound is tight up to a polylogarithmic factor. We also show that the lower bound is not asymptotically sharp as n is fixed and m⟶∞ .}, author = {Pach, János and Tardos, Gábor and Tóth, Géza}, booktitle = {28th International Symposium on Graph Drawing and Network Visualization}, isbn = {9783030687656}, issn = {1611-3349}, location = {Virtual, Online}, pages = {359--371}, publisher = {Springer Nature}, title = {{Crossings between non-homotopic edges}}, doi = {10.1007/978-3-030-68766-3_28}, volume = {12590}, year = {2020}, } @article{9308, author = {Avvakumov, Sergey and Wagner, Uli and Mabillard, Isaac and Skopenkov, A. B.}, issn = {0036-0279}, journal = {Russian Mathematical Surveys}, number = {6}, pages = {1156--1158}, publisher = {IOP Publishing}, title = {{Eliminating higher-multiplicity intersections, III. Codimension 2}}, doi = {10.1070/RM9943}, volume = {75}, year = {2020}, } @misc{9326, abstract = {The mitochondrial respiratory chain, formed by five protein complexes, utilizes energy from catabolic processes to synthesize ATP. Complex I, the first and the largest protein complex of the chain, harvests electrons from NADH to reduce quinone, while pumping protons across the mitochondrial membrane. Detailed knowledge of the working principle of such coupled charge-transfer processes remains, however, fragmentary due to bottlenecks in understanding redox-driven conformational transitions and their interplay with the hydrated proton pathways. Complex I from Thermus thermophilus encases 16 subunits with nine iron–sulfur clusters, reduced by electrons from NADH. Here, employing the latest crystal structure of T. thermophilus complex I, we have used microsecond-scale molecular dynamics simulations to study the chemo-mechanical coupling between redox changes of the iron–sulfur clusters and conformational transitions across complex I. First, we identify the redox switches within complex I, which allosterically couple the dynamics of the quinone binding pocket to the site of NADH reduction. Second, our free-energy calculations reveal that the affinity of the quinone, specifically menaquinone, for the binding-site is higher than that of its reduced, menaquinol forma design essential for menaquinol release. Remarkably, the barriers to diffusive menaquinone dynamics are lesser than that of the more ubiquitous ubiquinone, and the naphthoquinone headgroup of the former furnishes stronger binding interactions with the pocket, favoring menaquinone for charge transport in T. thermophilus. Our computations are consistent with experimentally validated mutations and hierarchize the key residues into three functional classes, identifying new mutation targets. Third, long-range hydrogen-bond networks connecting the quinone-binding site to the transmembrane subunits are found to be responsible for proton pumping. Put together, the simulations reveal the molecular design principles linking redox reactions to quinone turnover to proton translocation in complex I.}, author = {Gupta, Chitrak and Khaniya, Umesh and Chan, Chun and Dehez, Francois and Shekhar, Mrinal and Gunner, M. R. and Sazanov, Leonid A and Chipot, Christophe and Singharoy, Abhishek}, publisher = {American Chemical Society}, title = {{Charge transfer and chemo-mechanical coupling in respiratory complex I}}, doi = {10.1021/jacs.9b13450.s002}, year = {2020}, } @inproceedings{9415, abstract = {Optimizing convolutional neural networks for fast inference has recently become an extremely active area of research. One of the go-to solutions in this context is weight pruning, which aims to reduce computational and memory footprint by removing large subsets of the connections in a neural network. Surprisingly, much less attention has been given to exploiting sparsity in the activation maps, which tend to be naturally sparse in many settings thanks to the structure of rectified linear (ReLU) activation functions. In this paper, we present an in-depth analysis of methods for maximizing the sparsity of the activations in a trained neural network, and show that, when coupled with an efficient sparse-input convolution algorithm, we can leverage this sparsity for significant performance gains. To induce highly sparse activation maps without accuracy loss, we introduce a new regularization technique, coupled with a new threshold-based sparsification method based on a parameterized activation function called Forced-Activation-Threshold Rectified Linear Unit (FATReLU). We examine the impact of our methods on popular image classification models, showing that most architectures can adapt to significantly sparser activation maps without any accuracy loss. Our second contribution is showing that these these compression gains can be translated into inference speedups: we provide a new algorithm to enable fast convolution operations over networks with sparse activations, and show that it can enable significant speedups for end-to-end inference on a range of popular models on the large-scale ImageNet image classification task on modern Intel CPUs, with little or no retraining cost. }, author = {Kurtz, Mark and Kopinsky, Justin and Gelashvili, Rati and Matveev, Alexander and Carr, John and Goin, Michael and Leiserson, William and Moore, Sage and Nell, Bill and Shavit, Nir and Alistarh, Dan-Adrian}, booktitle = {37th International Conference on Machine Learning, ICML 2020}, issn = {2640-3498}, location = {Online}, pages = {5533--5543}, title = {{Inducing and exploiting activation sparsity for fast neural network inference}}, volume = {119}, year = {2020}, } @article{9526, abstract = {DNA methylation and histone H1 mediate transcriptional silencing of genes and transposable elements, but how they interact is unclear. In plants and animals with mosaic genomic methylation, functionally mysterious methylation is also common within constitutively active housekeeping genes. Here, we show that H1 is enriched in methylated sequences, including genes, of Arabidopsis thaliana, yet this enrichment is independent of DNA methylation. Loss of H1 disperses heterochromatin, globally alters nucleosome organization, and activates H1-bound genes, but only weakly de-represses transposable elements. However, H1 loss strongly activates transposable elements hypomethylated through mutation of DNA methyltransferase MET1. Hypomethylation of genes also activates antisense transcription, which is modestly enhanced by H1 loss. Our results demonstrate that H1 and DNA methylation jointly maintain transcriptional homeostasis by silencing transposable elements and aberrant intragenic transcripts. Such functionality plausibly explains why DNA methylation, a well-known mutagen, has been maintained within coding sequences of crucial plant and animal genes.}, author = {Choi, Jaemyung and Lyons, David B. and Kim, M. Yvonne and Moore, Jonathan D. and Zilberman, Daniel}, issn = {1097-4164}, journal = {Molecular Cell}, number = {2}, pages = {310--323.e7}, publisher = {Elsevier}, title = {{DNA methylation and histone H1 jointly repress transposable elements and aberrant intragenic transcripts}}, doi = {10.1016/j.molcel.2019.10.011}, volume = {77}, year = {2020}, } @article{9630, abstract = {Various kinds of data are routinely represented as discrete probability distributions. Examples include text documents summarized by histograms of word occurrences and images represented as histograms of oriented gradients. Viewing a discrete probability distribution as a point in the standard simplex of the appropriate dimension, we can understand collections of such objects in geometric and topological terms. Importantly, instead of using the standard Euclidean distance, we look into dissimilarity measures with information-theoretic justification, and we develop the theory needed for applying topological data analysis in this setting. In doing so, we emphasize constructions that enable the usage of existing computational topology software in this context.}, author = {Edelsbrunner, Herbert and Virk, Ziga and Wagner, Hubert}, issn = {1920180X}, journal = {Journal of Computational Geometry}, number = {2}, pages = {162--182}, publisher = {Carleton University}, title = {{Topological data analysis in information space}}, doi = {10.20382/jocg.v11i2a7}, volume = {11}, year = {2020}, } @inproceedings{9631, abstract = {The ability to leverage large-scale hardware parallelism has been one of the key enablers of the accelerated recent progress in machine learning. Consequently, there has been considerable effort invested into developing efficient parallel variants of classic machine learning algorithms. However, despite the wealth of knowledge on parallelization, some classic machine learning algorithms often prove hard to parallelize efficiently while maintaining convergence. In this paper, we focus on efficient parallel algorithms for the key machine learning task of inference on graphical models, in particular on the fundamental belief propagation algorithm. We address the challenge of efficiently parallelizing this classic paradigm by showing how to leverage scalable relaxed schedulers in this context. We present an extensive empirical study, showing that our approach outperforms previous parallel belief propagation implementations both in terms of scalability and in terms of wall-clock convergence time, on a range of practical applications.}, author = {Aksenov, Vitaly and Alistarh, Dan-Adrian and Korhonen, Janne}, booktitle = {Advances in Neural Information Processing Systems}, isbn = {9781713829546}, issn = {10495258}, location = {Vancouver, Canada}, pages = {22361--22372}, publisher = {Curran Associates}, title = {{Scalable belief propagation via relaxed scheduling}}, volume = {33}, year = {2020}, } @inproceedings{9632, abstract = {Second-order information, in the form of Hessian- or Inverse-Hessian-vector products, is a fundamental tool for solving optimization problems. Recently, there has been significant interest in utilizing this information in the context of deep neural networks; however, relatively little is known about the quality of existing approximations in this context. Our work examines this question, identifies issues with existing approaches, and proposes a method called WoodFisher to compute a faithful and efficient estimate of the inverse Hessian. Our main application is to neural network compression, where we build on the classic Optimal Brain Damage/Surgeon framework. We demonstrate that WoodFisher significantly outperforms popular state-of-the-art methods for oneshot pruning. Further, even when iterative, gradual pruning is allowed, our method results in a gain in test accuracy over the state-of-the-art approaches, for standard image classification datasets such as ImageNet ILSVRC. We examine how our method can be extended to take into account first-order information, as well as illustrate its ability to automatically set layer-wise pruning thresholds and perform compression in the limited-data regime. The code is available at the following link, https://github.com/IST-DASLab/WoodFisher.}, author = {Singh, Sidak Pal and Alistarh, Dan-Adrian}, booktitle = {Advances in Neural Information Processing Systems}, isbn = {9781713829546}, issn = {10495258}, location = {Vancouver, Canada}, pages = {18098--18109}, publisher = {Curran Associates}, title = {{WoodFisher: Efficient second-order approximation for neural network compression}}, volume = {33}, year = {2020}, } @inproceedings{9633, abstract = {The search for biologically faithful synaptic plasticity rules has resulted in a large body of models. They are usually inspired by – and fitted to – experimental data, but they rarely produce neural dynamics that serve complex functions. These failures suggest that current plasticity models are still under-constrained by existing data. Here, we present an alternative approach that uses meta-learning to discover plausible synaptic plasticity rules. Instead of experimental data, the rules are constrained by the functions they implement and the structure they are meant to produce. Briefly, we parameterize synaptic plasticity rules by a Volterra expansion and then use supervised learning methods (gradient descent or evolutionary strategies) to minimize a problem-dependent loss function that quantifies how effectively a candidate plasticity rule transforms an initially random network into one with the desired function. We first validate our approach by re-discovering previously described plasticity rules, starting at the single-neuron level and “Oja’s rule”, a simple Hebbian plasticity rule that captures the direction of most variability of inputs to a neuron (i.e., the first principal component). We expand the problem to the network level and ask the framework to find Oja’s rule together with an anti-Hebbian rule such that an initially random two-layer firing-rate network will recover several principal components of the input space after learning. Next, we move to networks of integrate-and-fire neurons with plastic inhibitory afferents. We train for rules that achieve a target firing rate by countering tuned excitation. Our algorithm discovers a specific subset of the manifold of rules that can solve this task. Our work is a proof of principle of an automated and unbiased approach to unveil synaptic plasticity rules that obey biological constraints and can solve complex functions.}, author = {Confavreux, Basile J and Zenke, Friedemann and Agnes, Everton J. and Lillicrap, Timothy and Vogels, Tim P}, booktitle = {Advances in Neural Information Processing Systems}, issn = {1049-5258}, location = {Vancouver, Canada}, pages = {16398--16408}, title = {{A meta-learning approach to (re)discover plasticity rules that carve a desired function into a neural network}}, volume = {33}, year = {2020}, } @unpublished{10012, abstract = {We prove that in the absence of topological changes, the notion of BV solutions to planar multiphase mean curvature flow does not allow for a mechanism for (unphysical) non-uniqueness. Our approach is based on the local structure of the energy landscape near a classical evolution by mean curvature. Mean curvature flow being the gradient flow of the surface energy functional, we develop a gradient-flow analogue of the notion of calibrations. Just like the existence of a calibration guarantees that one has reached a global minimum in the energy landscape, the existence of a "gradient flow calibration" ensures that the route of steepest descent in the energy landscape is unique and stable.}, author = {Fischer, Julian L and Hensel, Sebastian and Laux, Tim and Simon, Thilo}, booktitle = {arXiv}, title = {{The local structure of the energy landscape in multiphase mean curvature flow: weak-strong uniqueness and stability of evolutions}}, year = {2020}, } @unpublished{10022, abstract = {We consider finite-volume approximations of Fokker-Planck equations on bounded convex domains in R^d and study the corresponding gradient flow structures. We reprove the convergence of the discrete to continuous Fokker-Planck equation via the method of Evolutionary Γ-convergence, i.e., we pass to the limit at the level of the gradient flow structures, generalising the one-dimensional result obtained by Disser and Liero. The proof is of variational nature and relies on a Mosco convergence result for functionals in the discrete-to-continuum limit that is of independent interest. Our results apply to arbitrary regular meshes, even though the associated discrete transport distances may fail to converge to the Wasserstein distance in this generality.}, author = {Forkert, Dominik L and Maas, Jan and Portinale, Lorenzo}, booktitle = {arXiv}, pages = {33}, title = {{Evolutionary Γ-convergence of entropic gradient flow structures for Fokker-Planck equations in multiple dimensions}}, year = {2020}, } @inproceedings{10328, abstract = {We discus noise channels in coherent electro-optic up-conversion between microwave and optical fields, in particular due to optical heating. We also report on a novel configuration, which promises to be flexible and highly efficient.}, author = {Lambert, Nicholas J. and Mobassem, Sonia and Rueda Sanchez, Alfredo R and Schwefel, Harald G.L.}, booktitle = {OSA Quantum 2.0 Conference}, isbn = {9-781-5575-2820-9}, location = {Washington, DC, United States}, publisher = {Optica Publishing Group}, title = {{New designs and noise channels in electro-optic microwave to optical up-conversion}}, doi = {10.1364/QUANTUM.2020.QTu8A.1}, year = {2020}, } @inproceedings{10556, abstract = {In this paper, we present the first Asynchronous Distributed Key Generation (ADKG) algorithm which is also the first distributed key generation algorithm that can generate cryptographic keys with a dual (f,2f+1)-threshold (where f is the number of faulty parties). As a result, using our ADKG we remove the trusted setup assumption that the most scalable consensus algorithms make. In order to create a DKG with a dual (f,2f+1)- threshold we first answer in the affirmative the open question posed by Cachin et al. [7] on how to create an Asynchronous Verifiable Secret Sharing (AVSS) protocol with a reconstruction threshold of f+1