检索结果-内蒙古大学图书馆

Optimising Event-Driven Spiking Neural Network with Regularisation and Cutoff

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wu, Dengyu Jin, Gaojie Yu, Han Yi, Xinping Huang, Xiaowei Department of Computer Science University of Liverpool Liverpool United Kingdom National Mobile Communications Research Laboratory Southeast University Nanjing China State Key Laboratory of Computer Science Institute of Software CAS Beijing China Department of Electrical Engineering Chalmers University of Technology Gothenburg Sweden

Spiking neural network (SNN), as the next generation of artificial neural network (ANN), offer a closer mimicry of natural neural networks and hold promise for significant improvements in computational efficiency. However, the current SNN is trained to infer over a fixed duration, overlooking the potential of dynamic inference in SNN. In this paper, we strengthen the marriage between SNN and event-driven processing with a proposal to consider a cutoff in SNN, which can terminate SNN anytime during inference to achieve efficient inference. Two novel optimisation techniques are presented to achieve inference efficient SNN: a Top-K cutoff and a regularisation. The proposed regularisation influences the training process, optimising SNN for the cutoff, while the Top-K cutoff technique optimises the inference phase. We conduct an extensive set of experiments on multiple benchmark frame-based datasets, such as CIFAR10/100, Tiny-ImageNet, and event-based datasets, including CIFAR10-DVS, N-Caltech101 and DVS128 Gesture. The experimental results demonstrate the effectiveness of our techniques in both ANN-to-SNN conversion and direct training, enabling SNNs to require 1.76 to 2.76 × fewer timesteps for CIFAR-10, while achieving 1.64 to 1.95× fewer timesteps across all event-based datasets, with near-zero accuracy loss. These findings affirms the compatibility and potential benefits of our techniques in enhancing accuracy and reducing inference latency when integrated with existing methods. Code available: https://***/Dengyu-Wu/SNNCutoff © 2023, CC BY.

关键词： Computational efficiency

TIME-CRITICAL COMMUNICATION AND COMPUTATION FOR INTELLIGENT VEHICULAR NETWORKS

学校读者我要写书评

暂无评论

China Communications 2021年第6期18卷 I0004-I0006页

作者： Shanzhi Chen Tommy Svensson Sheng Zhou Shan Zhang State Key Laboratory of Wireless Mobile Communications CATT Chalmers University of Technology in Gothen-burg Sweden Department of Electronic Engineering Tsinghua University School of Computer Science and Engineering Beihang University BeijingChina

Vehicular networks are expected to empower auto mated driving and intelligent transportation via vehicle-to-everything(V2X)communications and edge/cloud-assisted computation,and in the meantime Cellular V2X(C-V2X)is gaining wide support from the global industrial *** 5G NR-V2X technology is the evolution of LTE-V2X,which is expected to provide ultra-Reliable and Low-Latency Communications(uRLLC)with 1ms latency and 99.999%***,vehicular networks still face great challenges in supporting many emerging time-critical applications,which comprise sensing,communication and computation as closed-loops.

关键词： networks V2X driving

Opening the Black Box: How Boolean AI can Support Legal Analysis

学校读者我要写书评

暂无评论

Opening the Black Box: How Boolean AI can Support Legal Anal...

computer Communication and Artificial Intelligence (CCAI), International Conference on

作者： Grazia Garzo Stefano Ribes Alessandro Palumbo Department of Social Political and Cognitives Sciences University of Siena Siena Italy Department of Computer Science and Engineering Chalmers University of Technology Göteborg Sweden CentraleSupélec. Inria Univ Rennes CNRS IRISA Cesson-Sévigné France

ISBN: (数字)9798350362763

ISBN: (纸本)9798350362770

In crime scene scenarios, there are various factors to consider when determining a suspect's guilt. However, the process of extracting and assessing these factors can be time-consuming, often taking years and incurring significant legal expenses. Judges are now exploring the potential of artificial intelligence techniques and machine learning computations within the justice system. Specifically, in the realm of criminal justice, these methodologies have the potential to aid in investigations and decision-making processes. Utilizing machine learning approaches can thus expedite the bureaucratic process, potentially making it more efficient. We introduce an idea of an approach that could provide fast and explainable support in the evaluation of guilt. Our approach relies on computations based on the presence or absence of 44 features describing the crime scene. Then, by a boolean function, we determined the final verdict of the legal case (only a subset of the extracted features are relevant to evaluate the guilt prediction). To demonstrate the practicality of our proposal, we conducted experiments based on 79 road homicide cases in Italy. As a consequence, the boolean evaluation was done according to Italian law principles. With our system, we reached a 83.2 % accuracy rate in extracting features from the legal ruling texts and a 69.6% accuracy in guilt prediction.

关键词： Analytical models Accuracy Boolean functions Law Roads Decision making Machine learning

Joint Communications and Sensing Design for Multi-Carrier MIMO Systems

学校读者我要写书评

暂无评论

Joint Communications and Sensing Design for Multi-Carrier MI...

IEEE/SP Workshop on Statistical Signal Processing (SSP)

作者： Nhan Thanh Nguyen Nir Shlezinger Khac-Hoang Ngo Van-Dinh Nguyen Markku Juntti Centre for Wireless Communications University of Oulu Finland School of ECE Ben-Gurion University of the Negev Beer-Sheva Israel Department of Electrical Engineering Chalmers University of Technology Gothenburg Sweden College of Engineering and Computer Science VinUniversity Hanoi Vietnam

In conventional joint communications and sensing (JCAS) designs for multi-carrier multiple-input multiple-output (MIMO) systems, the dual-functional waveforms are often optimized for the whole frequency band, resulting in limited communications–sensing performance tradeoff. To overcome the limitation, we propose employing a subset of subcarriers for JCAS, while the communications function is performed over all the subcarriers. This offers more degrees of freedom to enhance the communications performance under a given sensing accuracy. We first formulate the rate maximization under the sensing accuracy constraint to optimize the beamformers and JCAS subcarriers. The problem is solved via Riemannian manifold optimization and closed-form solutions. Numerical results for an 8 × 4 MIMO system with 64 subcarriers show that compared to the conventional subcarrier sharing scheme, the proposed scheme employing 16 JCAS subcarriers offers 60% improvement in the achievable communications rate at the signal-to-noise ratio of 10 dB. Meanwhile, this scheme generates the sensing beampattern with the same quality as the conventional JCAS design.

关键词：

Diverse Data Expansion with Semi-Supervised k-Determinantal Point Processes

学校读者我要写书评

暂无评论

Diverse Data Expansion with Semi-Supervised k-Determinantal ...

IEEE International Conference on Big Data

作者： Simon Johansson Ola Engkvist Morteza Haghir Chehreghani Alexander Schliep Department of Computer Science and Engineering Chalmers University of Technology Gothenburg Sweden Molecular AI Discovery Sciences R&D AstraZeneca Gothenburg Sweden Faculty of Health Sciences Brandenburg University of Technology Cottbus-Senftenberg Cottbus Germany

Determinantal point processes (DPPs) have become prominent in data summarization and recommender system tasks for their ability to simultaneously model diversity as well as relevance. In practical applications, k-Determinantal point processes (k-DPPs) are used to yield a selection of k items from a set of size N that are the most representative of the set. In this paper, we study a special case of the diverse subset selection problem where a fixed set GO is already given as a forced recommendation and the task is to determine the remainder of the recommendation G1. The standard k-DPP optimization objectives here can suggest items that are close to optimal when considering only items in G1, but are arbitrarily close to items in G0, i.e., they might not be sufficiently diverse w.r.t. G0. We explore a semi-supervised k-DPP objective that simultaneously considers G0 and G1 and compares the difference between the two recommendations. We demonstrate our findings using multiple examples where the diverse subset selection problem with forced recommendation is important in practice.

关键词：

ERROR ANALYSIS FOR DISCONTINUOUS GALERKIN TIME-STEPPING METHODS FOR NONLINEAR PARABOLIC EQUATIONS VIA MAXIMAL REGULARITY

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Akrivis, Georgios Larsson, Stig Department of Computer Science and Engineering University of Ioannina Ioannina451 10 Greece Institute of Applied and Computational Mathematics FORTH Heraklion Crete700 13 Greece Department of Mathematical Sciences Chalmers University of Technology University of Gothenburg GothenburgSE–412 96 Sweden

We consider the discretization of a class of nonlinear parabolic equations by discontinuous Galerkin time-stepping methods and establish a priori as well as conditional a posteriori error estimates. Our approach is motivated by the error analysis in [9] for Runge–Kutta methods for nonlinear parabolic equations;in analogy to [9], the proofs are based on maximal regularity properties of discontinuous Galerkin methods for nonautonomous linear parabolic *** Codes 65M12, 65M15, 65M60 © 2024, CC BY.

关键词： Galerkin methods

Autonomous Drug Design with Multi-Armed Bandits

学校读者我要写书评

暂无评论

Autonomous Drug Design with Multi-Armed Bandits

2022 IEEE International Conference on Big Data, Big Data 2022

作者： Svensson, Hampus Gummesson Jannik Bjerrum, Esben Tyrchan, Christian Engkvist, Ola Chehreghani, Morteza Haghir Chalmers University of Technology Department of Computer Science and Engineering Gothenburg Sweden Discovery Sciences R and D AstraZeneca Molecular Ai Gothenburg Sweden BioPharmaceuticals R and D AstraZeneca Medicinal Chemistry Research and Early Development Gothenburg Sweden Odyssey Therapeutics CambridgeMA United States

ISBN: (纸本)9781665480451

Recent developments in artificial intelligence and automation support a new drug design paradigm: autonomous drug design. Under this paradigm, generative models can provide suggestions on thousands of molecules with specific properties, and automated laboratories can potentially make, test and analyze molecules with minimal human supervision. However, since still only a limited number of molecules can be synthesized and tested, an obvious challenge is how to efficiently select among provided suggestions in a closed-loop system. We formulate this task as a stochastic multi-armed bandit problem with multiple plays, volatile arms and similarity information. To solve this task, we adapt previous work on multi-armed bandits to this setting, and compare our solution with random sampling, greedy selection and decaying-epsilon-greedy selection strategies. According to our simulation results, our approach has the potential to perform better exploration and exploitation of the chemical space for autonomous drug design. © 2022 IEEE.

关键词： Molecules

A THEORY OF INITIALISATION’S IMPACT ON SPECIALISATION

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Jarvis, Devon Lee, Sebastian Dominé, Clémentine Carla Juliette Saxe, Andrew M. Mannelli, Stefano Sarao School of Computer Science and Applied Mathematics University of the Witwatersrand South Africa Center for Computational Neuroscience Flatiron Institute Simons Foundation Gatsby Computational Neuroscience Unit Sainsbury Wellcome Centre UCL United Kingdom Data Science and AI Computer Science and Engineering Chalmers University of Technology University of Gothenburg Sweden Machine Intelligence and Neural Discovery Institute University of the Witwatersrand South Africa CIFAR Azrieli Global Scholar CIFAR

Prior work has demonstrated a consistent tendency in neural networks engaged in continual learning tasks, wherein intermediate task similarity results in the highest levels of catastrophic interference. This phenomenon is attributed to the network’s tendency to reuse learned features across tasks. However, this explanation heavily relies on the premise that neuron specialisation occurs, i.e. the emergence of localised representations. Our investigation challenges the validity of this assumption. Using theoretical frameworks for the analysis of neural networks, we show a strong dependence of specialisation on the initial condition. More precisely, we show that weight imbalance and high weight entropy can favour specialised solutions. We then apply these insights in the context of continual learning, first showing the emergence of a monotonic relation between task-similarity and forgetting in non-specialised networks. Finally, we show that specialization by weight imbalance is beneficial on the commonly employed elastic weight consolidation regularisation technique. Copyright © 2025, The Authors. All rights reserved.

关键词： Neural networks

Unsourced Multiple Access With Random User Activity

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Ngo, Khac-Hoang Lancho, Alejandro Durisi, Giuseppe Amat, Alexandre Graell i The Department of Electrical Engineering Chalmers University of Technology Gothenburg41296 Sweden The Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology CambridgeMA02139 United States

To account for the massive uncoordinated random access scenario, which is relevant for the Internet of Things, Polyanskiy (2017) proposed a novel formulation of the multiple-access problem, commonly referred to as unsourced multiple access, where all users employ a common codebook and the receiver decodes up to a permutation of the messages. In this paper, we extend this seminal work to the case where the number of active users is random and unknown a priori. We define a random-access code accounting for both misdetection (MD) and false alarm (FA), and derive a random-coding achievability bound for the Gaussian multiple access channel. Our bound captures the fundamental trade-off between MD and FA probabilities. It suggests that the lack of knowledge of the number of active users entails a small penalty in energy efficiency when the target MD and FA probabilities are high. However, as the target MD and FA probabilities decrease, the energy efficiency penalty becomes more significant. For example, in a typical IoT scenario with framelength 19200 complex channel uses and 25-300 active users in average, the required energy per bit to achieve both MD and FA probabilities below 10−1, predicted by our bound, is only 0.5-0.7 dB higher than that predicted by the bound in Polyanskiy (2017) for a known number of active users. This gap increases to 3-4 dB when the target MD probability and/or FA probability is below 10−3. Taking both MD and FA into account, we use our bound to benchmark the energy efficiency of slotted-ALOHA with multi-packet reception, of a decoder that simply treats interference as noise, and of some recently proposed unsourced multiple access schemes. Numerical results suggest that, when the target MD and FA probabilities are high, it is effective to estimate the number of active users, then treat this estimate as the true value, and use a coding scheme that performs well for the case of known number of active users. However, this approach becomes energy in

关键词： Energy efficiency