查看更多>>摘要：In this work, we demonstrate provable guarantees on the training of a single ReLU gate in hitherto unexplored regimes. We give a simple iterative stochastic algorithm that can train a ReLU gate in the realizable setting in linear time while using significantly milder conditions on the data distribution than previous such results.& nbsp;Leveraging certain additional moment assumptions, we also show a first-of-its-kind approximate recovery of the true label generating parameters under an (online) data-poisoning attack on the true labels, while training a ReLU gate by the same algorithm. Our guarantee is shown to be nearly optimal in the worst case and its accuracy of recovering the true weight degrades gracefully with increasing probability of attack and its magnitude.& nbsp;For both the realizable and the non-realizable cases as outlined above, our analysis allows for minibatching and computes how the convergence time scales with the mini-batch size. We corroborate our theorems with simulation results which also bring to light a striking similarity in trajectories between our algorithm and the popular S.G.D. algorithm - for which similar guarantees as here are still unknown. (C) 2022 Elsevier Ltd. All rights reserved.
查看更多>>摘要：Despite the well-recognized role of the posterior parietal cortex (PPC) in processing sensory information to guide action, the differential encoding properties of this dynamic processing, as operated by different PPC brain areas, are scarcely known. Within the monkey's PPC, the superior parietal lobule hosts areas V6A, PEc, and PE included in the dorso-medial visual stream that is specialized in planning and guiding reaching movements. Here, a Convolutional Neural Network (CNN) approach is used to investigate how the information is processed in these areas. We trained two macaque monkeys to perform a delayed reaching task towards 9 positions (distributed on 3 different depth and direction levels) in the 3D peripersonal space. The activity of single cells was recorded from V6A, PEc, PE and fed to convolutional neural networks that were designed and trained to exploit the temporal structure of neuronal activation patterns, to decode the target positions reached by the monkey. Bayesian Optimization was used to define the main CNN hyper-parameters. In addition to discrete positions in space, we used the same network architecture to decode plausible reaching trajectories. We found that data from the most caudal V6A and PEc areas outperformed PE area in the spatial position decoding. In all areas, decoding accuracies started to increase at the time the target to reach was instructed to the monkey, and reached a plateau at movement onset. The results support a dynamic encoding of the different phases and properties of the reaching movement differentially distributed over a network of interconnected areas. This study highlights the usefulness of neurons' firing rate decoding via CNNs to improve our understanding of how sensorimotor information is encoded in PPC to perform reaching movements. The obtained results may have implications in the perspective of novel neuroprosthetic devices based on the decoding of these rich signals for faithfully carrying out patient's intentions.(C) 2022 Published by Elsevier Ltd.
查看更多>>摘要：Over the last 10 to 15 years, active inference has helped to explain various brain mechanisms from habit formation to dopaminergic discharge and even modelling curiosity. However, the current implementations suffer from an exponential (space and time) complexity class when computing the prior over all the possible policies up to the time-horizon. Fountas et al. (2020) used Monte Carlo tree search to address this problem, leading to impressive results in two different tasks. In this paper, we present an alternative framework that aims to unify tree search and active inference by casting planning as a structure learning problem. Two tree search algorithms are then presented. The first propagates the expected free energy forward in time (i.e., towards the leaves), while the second propagates it backward (i.e., towards the root). Then, we demonstrate that forward and backward propagations are related to active inference and sophisticated inference, respectively, thereby clarifying the differences between those two planning strategies. (c) 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
查看更多>>摘要：In building artificial intelligence (AI) agents, referring to how brains function in real environments can accelerate development by reducing the design space. In this study, we propose a probabilistic generative model (PGM) for navigation in uncertain environments by integrating the neuroscientific knowledge of hippocampal formation (HF) and the engineering knowledge in robotics and AI, namely, simultaneous localization and mapping (SLAM). We follow the approach of brain reference architecture (BRA) (Yamakawa, 2021) to compose the PGM and outline how to verify the model. To this end, we survey and discuss the relationship between the HF findings and SLAM models. The proposed hippocampal formation-inspired probabilistic generative model (HF-PGM) is designed to be highly consistent with the anatomical structure and functions of the HF. By referencing the brain, we elaborate on the importance of integration of egocentric/allocentric information from the entorhinal cortex to the hippocampus and the use of discrete-event queues. (c) 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
查看更多>>摘要：This paper investigates vehicle trajectory prediction problems in real traffic scenarios by fully harnessing the spatio-temporal dependencies between multiple vehicles. The existing GCN-based trajectory predictions are often considered in a single traffic scene without time attributes, complete interaction information, dynamic graph-based model, etc. Time and interaction aware models are more challenging than the existing ones. Despite very well does the graph-based model describe the relationship between driving vehicles, the critical problem in the traffic scene is how to deeply explore the spatiotemporal characteristics of dynamic graphs. Therefore, a novel dynamic graph and interaction-aware neural network model called as DGInet is proposed by combining a semi-global graph mechanism and an M-product based graph convolutional network, which are built into novel dual-network architecture in the entire model. The DGInet is built by exploiting the dynamic interaction in depth between driving vehicles in urban traffic scenarios, and then realized by utilizing semi-global graph convolution operations on the input data cell to capture the basic spatial interaction features of the driving scene. Meanwhile, the dynamic graph is further extracted by a novel M-product approach, in which the embedding of the model is then established along with the embedding of the semi-global network to perform the final embedding. Extensive experiments have been conducted on the two public datasets, named NGSIM and Apollo respectively, to show that our approach outperforms the existing ones with better performance and less computing time. Besides the real-world Shenzhen traffic dataset, China, is also developed to verify the effectiveness of our approach. (C) 2022 Elsevier Ltd. All rights reserved.
查看更多>>摘要：Categorical relationships between objects are encoded as overlapped neural representations in the brain, where the more similar the objects are, the larger the correlations between their evoked neuronal responses. These representation correlations, however, inevitably incur interference when memories are retrieved. Here, we propose that neural feedback, which is widely observed in the brain but whose function remains largely unknown, contributes to disentangle neural correlations to improve information retrieval. We study a hierarchical neural network storing the hierarchical categorical information of objects, and information retrieval goes from rough-to-fine, aided by the push-pull neural feedback. We elucidate that the push and the pull components of the feedback suppress the interferences due to the representation correlations between objects from different and the same categories, respectively. Our model reproduces the push-pull phenomenon observed in neural data and sheds light on our understanding of the role of feedback in neural information processing. (C)& nbsp;2022 The Authors. Published by Elsevier Ltd.& nbsp;
查看更多>>摘要：Conversational gestures have a crucial role in realizing natural interactions with virtual agents and robots. Data-driven approaches, such as deep learning and machine learning, are promising in constructing the gesture generation model, which automatically provides the gesture motion for speech or spoken texts. This study experimentally analyzes a deep learning-based gesture generation model from spoken text using a convolutional neural network. The proposed model takes a sequence of spoken words as the input and outputs a sequence of 2D joint coordinates representing the conversational gesture motion. We prepare a dataset consisting of gesture motions and spoken texts by adding text information to an existing dataset and train the models using specific speaker's data. The quality of the generated gestures is compared with those from an existing speech-to-gesture generation model through a user perceptual study. The subjective evaluation shows that the model performance is comparable or superior to those by the existing speech-to-gesture generation model. In addition, we investigate the importance of data cleansing and loss function selection in the text-to-gesture generation model. We further examine the model transferability between speakers. The experimental results demonstrate successful model transferability of the proposed model. Finally, we show that the text-to-gesture generation model can produce good quality gestures even when using a transformer architecture.(c) 2022 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
查看更多>>摘要：An elegant quantum version of least-square support vector machine, which is exponentially faster than the classical counterpart, was given by Rebentrost et al. using the matrix inversion algorithm (HHL). However, the application of the HHL algorithm is restricted when the structure of the input matrix is not well. The iteration algorithms such as the Newton method are widespread in training the classical support vector machine. This paper demonstrates a quantum support vector machine based on the regularized Newton method (RN-QSVM), which achieves an exponential speed-up over classical algorithm. At first, the regularized quantum Newton algorithm is proposed to get rid of the constraint of input matrix. Then we train the RN-QSVM by using the regularized quantum Newton algorithm and classify a query sample by constructing the quantum state. Experiments demonstrate that RNQSVM respectively provides advantages in terms of accuracy, robustness, and complexity compared to QSLS-SVM, LS-QSVM, and the classical method.
查看更多>>摘要：This paper is dedicated to solving the k-winners-take-all problem with large-scale input signals in a distributed manner. According to the decomposition of global input signals, a novel dynamical system consisting of multiple coordinated neural networks is proposed for finding the k largest inputs. In the system, each neural network is designed to tackle its available partial inputs only for a local objective ki (ki <=& nbsp;k). Simultaneously, a consensus-based approach is adopted to coordinate multiple neural networks for achieving the global objective k. In addition, an inertial term is introduced in each neural network for regulating its transient behavior, which has the potential of accelerating the convergence. By developing a cocoercive operator, we theoretically prove that the multiple neural networks with inertial terms converge asymptotically/exponentially to the k-winners-take-all solution exactly from arbitrary initial states for whatever decomposition of inputs and objective. Furthermore, some extensions to distributed constrained k-winners-take-all are also investigated. Finally, simulation results are presented to substantiate the effectiveness of the proposed system as well as its superior performance over existing distributed networks. (C)& nbsp;2022 Elsevier Ltd. All rights reserved.
NSTL
Elsevier