查看更多>>摘要：? 2021 Elsevier LtdThis work investigates the stability and dissipativity problems for neural networks with time-varying delay. By the construction of new augmented Lyapunov–Krasovskii functionals based on integral inequality and the use of zero equality approach, three improved results are proposed in the forms of linear matrix inequalities. And, based on the stability results, the dissipativity analysis for NNs with time-varying delays was investigated. Through some numerical examples, the superiority and effectiveness of the proposed results are shown by comparing the existing works.
查看更多>>摘要：? 2021 Elsevier LtdDeep neural networks unlocked a vast range of new applications by solving tasks of which many were previously deemed as reserved to higher human intelligence. One of the developments enabling this success was a boost in computing power provided by special purpose hardware, such as graphic or tensor processing units. However, these do not leverage fundamental features of neural networks like parallelism and analog state variables. Instead, they emulate neural networks relying on binary computing, which results in unsustainable energy consumption and comparatively low speed. Fully parallel and analogue hardware promises to overcome these challenges, yet the impact of analogue neuron noise and its propagation, i.e. accumulation, threatens rendering such approaches inept. Here, we determine for the first time the propagation of noise in deep neural networks comprising noisy nonlinear neurons in trained fully connected layers. We study additive and multiplicative as well as correlated and uncorrelated noise, and develop analytical methods that predict the noise level in any layer of symmetric deep neural networks or deep neural networks trained with back propagation. We find that noise accumulation is generally bound, and adding additional network layers does not worsen the signal to noise ratio beyond a limit. Most importantly, noise accumulation can be suppressed entirely when neuron activation functions have a slope smaller than unity. We therefore developed the framework for noise in fully connected deep neural networks implemented in analog systems, and identify criteria allowing engineers to design noise-resilient novel neural network hardware.
查看更多>>摘要：? 2021 Elsevier LtdBased on the theories of inertial systems, a second-order accelerated neurodynamic approach is designed to solve a distributed convex optimization with inequality and set constraints. Most of the existing approaches for distributed convex optimization problems are usually first-order ones, and it is usually hard to analyze the convergence rate for the state solution of those first-order approaches. Due to the control design for the acceleration, the second-order neurodynamic approaches can often achieve faster convergence rate. Moreover, the existing second-order approaches are mostly designed to solve unconstrained distributed convex optimization problems, and are not suitable for solving constrained distributed convex optimization problems. It is acquired that the state solution of the designed neurodynamic approach in this paper converges to the optimal solution of the considered distributed convex optimization problem. An error function which demonstrates the performance of the designed neurodynamic approach, has a superquadratic convergence. Several numerical examples are provided to show the effectiveness of the presented second-order accelerated neurodynamic approach.
查看更多>>摘要：? 2021 The Author(s)Graph construction plays an essential role in graph-based label propagation since graphs give some information on the structure of the data manifold. While most graph construction methods rely on predefined distance calculation, recent algorithms merge the task of label propagation and graph construction in a single process. Moreover, the use of several descriptors is proved to outperform a single descriptor in representing the relation between the nodes. In this article, we propose a Multiple-View Consistent Graph construction and Label propagation algorithm (MVCGL) that simultaneously constructs a consistent graph based on several descriptors and performs label propagation over unlabeled samples. Furthermore, it provides a mapping function from the feature space to the label space with which we estimate the label of unseen samples via a linear projection. The constructed graph does not rely on a predefined similarity function and exploits data and label smoothness. Experiments conducted on three face and one handwritten digit databases show that the proposed method can gain better performance compared to other graph construction and label propagation methods.
查看更多>>摘要：? 2021 The Author(s)In this work, we introduce, justify and demonstrate the Corrective Source Term Approach (CoSTA)—a novel approach to Hybrid Analysis and Modeling (HAM). The objective of HAM is to combine physics-based modeling (PBM) and data-driven modeling (DDM) to create generalizable, trustworthy, accurate, computationally efficient and self-evolving models. CoSTA achieves this objective by augmenting the governing equation of a PBM model with a corrective source term generated using a deep neural network. In a series of numerical experiments on one-dimensional heat diffusion, CoSTA is found to outperform comparable DDM and PBM models in terms of accuracy – often reducing predictive errors by several orders of magnitude – while also generalizing better than pure DDM. Due to its flexible but solid theoretical foundation, CoSTA provides a modular framework for leveraging novel developments within both PBM and DDM. Its theoretical foundation also ensures that CoSTA can be used to model any system governed by (deterministic) partial differential equations. Moreover, CoSTA facilitates interpretation of the DNN-generated source term within the context of PBM, which results in improved explainability of the DNN. These factors make CoSTA a potential door-opener for data-driven techniques to enter high-stakes applications previously reserved for pure PBM.
查看更多>>摘要：? 2021 Elsevier LtdDespite significant improvements in contemporary machine learning, symbolic methods currently outperform artificial neural networks on tasks that involve compositional reasoning, such as goal-directed planning and logical inference. This illustrates a computational explanatory gap between cognitive and neurocomputational algorithms that obscures the neurobiological mechanisms underlying cognition and impedes progress toward human-level artificial intelligence. Because of the strong relationship between cognition and working memory control, we suggest that the cognitive abilities of contemporary neural networks are limited by biologically-implausible working memory systems that rely on persistent activity maintenance and/or temporal nonlocality. Here we present NeuroLISP, an attractor neural network that can represent and execute programs written in the LISP programming language. Unlike previous approaches to high-level programming with neural networks, NeuroLISP features a temporally-local working memory based on itinerant attractor dynamics, top-down gating, and fast associative learning, and implements several high-level programming constructs such as compositional data structures, scoped variable binding, and the ability to manipulate and execute programmatic expressions in working memory (i.e., programs can be treated as data). Our computational experiments demonstrate the correctness of the NeuroLISP interpreter, and show that it can learn non-trivial programs that manipulate complex derived data structures (multiway trees), perform compositional string manipulation operations (PCFG SET task), and implement high-level symbolic AI algorithms (first-order unification). We conclude that NeuroLISP is an effective neurocognitive controller that can replace the symbolic components of hybrid models, and serves as a proof of concept for further development of high-level symbolic programming in neural networks.
查看更多>>摘要：? 2021 The Author(s)In this paper, a novel data-driven method for weight initialization of Multilayer Perceptrons and Convolutional Neural Networks based on discriminant learning is proposed. The approach relaxes some of the limitations of competing data-driven methods, including unimodality assumptions, limitations on the architectures related to limited maximal dimensionalities of the corresponding projection spaces, as well as limitations related to high computational requirements due to the need of eigendecomposition on high-dimensional data. We also consider assumptions of the method on the data and propose a way to account for them in a form of a new normalization layer. The experiments on three large-scale image datasets show improved accuracy of the trained models compared to competing random-based and data-driven weight initialization methods, as well as better convergence properties in certain cases.
查看更多>>摘要：? 2021 Elsevier LtdLOBSTER (LOss-Based SensiTivity rEgulaRization) is a method for training neural networks having a sparse topology. Let the sensitivity of a network parameter be the variation of the loss function with respect to the variation of the parameter. Parameters with low sensitivity, i.e. having little impact on the loss when perturbed, are shrunk and then pruned to sparsify the network. Our method allows to train a network from scratch, i.e. without preliminary learning or rewinding. Experiments on multiple architectures and datasets show competitive compression ratios with minimal computational overhead.
查看更多>>摘要：? 2021 Elsevier LtdThis work addresses the problem of cross-modality inference (CMI), i.e., inferring missing data of unavailable perceptual modalities (e.g., sound) using data from available perceptual modalities (e.g., image). We overview single-modality variational autoencoder methods and discuss three problems of computational cross-modality inference, arising from recent developments in multimodal generative models. Inspired by neural mechanisms of human recognition, we contribute the NEXUS model, a novel hierarchical generative model that can learn a multimodal representation of an arbitrary number of modalities in an unsupervised way. By exploiting hierarchical representation levels, NEXUS is able to generate high-quality, coherent data of missing modalities given any subset of available modalities. To evaluate CMI in a natural scenario with a high number of modalities, we contribute the “Multimodal Handwritten Digit” (MHD) dataset, a novel benchmark dataset that combines image, motion, sound and label information from digit handwriting. We access the key role of hierarchy in enabling high-quality samples during cross-modality inference and discuss how a novel training scheme enables NEXUS to learn a multimodal representation robust to missing modalities at test time. Our results show that NEXUS outperforms current state-of-the-art multimodal generative models in regards to their cross-modality inference capabilities.
查看更多>>摘要：? 2021 Elsevier LtdHuman adaptive behavior requires continually learning and performing a wide variety of tasks, often with very little practice. To accomplish this, it is crucial to separate neural representations of different tasks in order to avoid interference. At the same time, sharing neural representations supports generalization and allows faster learning. Therefore, a crucial challenge is to find an optimal balance between shared versus separated representations. Typically, models of human cognition employ top-down modulatory signals to separate task representations, but there exist surprisingly little systematic computational investigations of how such modulation is best implemented. We identify and systematically evaluate two crucial features of modulatory signals. First, top-down input can be processed in an additive or multiplicative manner. Second, the modulatory signals can be adaptive (learned) or non-adaptive (random). We cross these two features, resulting in four modulation networks which are tested on a variety of input datasets and tasks with different degrees of stimulus-action mapping overlap. The multiplicative adaptive modulation network outperforms all other networks in terms of accuracy. Moreover, this network develops hidden units that optimally share representations between tasks. Specifically, different than the binary approach of currently popular latent state models, it exploits partial overlap between tasks.
NSTL
Elsevier