检索结果-内蒙古大学图书馆

Uncovering dynamical equations of stochastic decision models using data-driven SINDy algorithm

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lenfesty, Brendan Bhattacharyya, Saugat Wong-Lin, KongFatt Intelligent Systems Research Centre School of Computing Engineering and Intelligent Systems Ulster University Magee campus Derry Northern Ireland Londonderry United Kingdom

Decision formation in perceptual decision-making involves sensory evidence accumulation instantiated by the temporal integration of an internal decision variable towards some decision criterion or threshold, as described by sequential sampling theoretical models. The decision variable can be represented in the form of experimentally observable neural activities. Hence, elucidating the appropriate theoretical model becomes crucial to understanding the mechanisms underlying perceptual decision formation. Existing computational methods are limited to either fitting of choice behavioural data or linear model estimation from neural activity data. In this work, we made use of sparse identification of nonlinear dynamics (SINDy), a data-driven approach, to elucidate the deterministic linear and nonlinear components of often-used stochastic decision models within reaction time task paradigms. Based on the simulated decision variable activities of the models, SINDy, enhanced with approaches using multiple trials, could readily estimate the dynamical equations, choice accuracy and decision time of the models across a range of signal-to-noise ratio values. In particular, SINDy performed the best using the more memory intensive multi-trial approach while trial-averaging of parameters performed more moderately. The single-trial approach, although expectedly not performing as well, may be useful for real-time modelling. Taken together, our work offers alternative approaches for SINDy to uncover the dynamics in perceptual decision-making, and more generally, for first-passage time problems. © 2024, CC BY.

关键词： Stochastic models

Can multivariate Granger causality detect directed connectivity of a multistable and dynamic biological decision network model?

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Asadpour, Abdoreza Wong-Lin, KongFatt Intelligent Systems Research Centre School of Computing Engineering and Intelligent Systems Ulster University Magee campus Northern Ireland Derry-Londonderry United Kingdom

Extracting causal connections can advance interpretable AI and machine learning. Granger causality (GC) is a robust statistical method for estimating directed influences (DC) between signals. While GC has been widely applied to analysing neuronal signals in biological neural networks and other domains, its application to complex, nonlinear, and multistable neural networks is less explored. In this study, we applied time-domain multivariate Granger causality (MVGC) to the time series neural activity of all nodes in a trained multistable biologically based decision neural network model with real-time decision uncertainty monitoring. Our analysis demonstrated that challenging two-choice decisions, where input signals could be closely matched, and the appropriate application of fine-grained sliding time windows, could readily reveal the original model’s DC. Furthermore, the identified DC varied based on whether the network had correct or error decisions. Integrating the identified DC from different decision outcomes recovered most of the original model’s architecture, despite some spurious and missing connectivity. This approach could be used as an initial exploration to enhance the interpretability and transparency of dynamic multistable and nonlinear biological or AI systems by revealing causal connections throughout different phases of neural network dynamics and outcomes. © 2024, CC BY-NC-SA.

关键词： Neural network models

Promise and Peril of Collaborative Code Generation Models: Balancing Effectiveness and Memorization

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Chen, Zhi Jiang, Lingxiao Centre for Research on Intelligent Software Engineering School of Computing and Information Systems Singapore Management University Singapore

In the rapidly evolving field of machine learning, training models with datasets from various locations and organizations presents significant challenges due to privacy and legal concerns. The exploration of effective collaborative training settings, which are capable of leveraging valuable knowledge from distributed and isolated datasets, is increasingly *** study investigates key factors that impact the effectiveness of collaborative training methods in code next-token prediction, as well as the correctness and utility of the generated code, showing the promise of such methods. Additionally, we evaluate the memorization of different participant training data across various collaborative training settings, including centralized, federated, and incremental training, showing their potential risks in leaking data. Our findings indicate that the size and diversity of code datasets are pivotal factors influencing the success of collaborative trained code models. We demonstrate that federated learning achieves competitive performance compared to centralized training while offering better data protection, as evidenced by lower memorization ratios in the generated code. However, federated learning can still produce verbatim code snippets from hidden training data, potentially violating data privacy or copyright. Our study further explores the patterns of effectiveness and memorization in incremental learning, emphasizing the importance of the sequence in which individual participant datasets are introduced. Also, we identify the memorization phenomenon of cross-organizational clones as a prevalent challenge in both centralized and federated learning scenarios. Our findings highlight the persistent risk of data leakage during inference, even when training data remains unseen. We conclude with strategic recommendations for practitioners and researchers to optimize the use of multisource datasets, thereby propelling the cross-organizational collaboration forward. Cop

关键词： Collaborative learning

Research on Target Localization of Automatic Picking Robot for Rock Sugar Orange 23

学校读者我要写书评

暂无评论

Research on Target Localization of Automatic Picking Robot f...

2023 International Conference on Computer, Vision and intelligent Technology, ICCVIT 2023

作者： Guo, Yuchen Liu, Dong Li, Qingbo Jiang, Hang Peng, Taomo Zeng, Xiongjian Liu, Ailin School of Computer and Artificial Intelligence Xiangnan University Chenzhou423000 China Hunan Engineering Research Center of Advanced Embedded Computing and Intelligent Medical Systems Xiangnan University Chenzhou423000 China

ISBN: (纸本)9798400708701

In the production process of rock sugar oranges, picking is the most labor-intensive step. intelligent replacement of traditional manual labor, timely and rapid picking of rock sugar oranges, improving fruit picking quality, is of great significance for expanding the rock sugar orange market. How to effectively recognize and locate rock sugar oranges is a key technology in machine vision in the research of automatic picking robots. A rock sugar orange positioning method is proposed. Firstly, the rock sugar orange image is preprocessed, binarized using Otsu adaptive threshold segmentation, and the positioning parameters and radius of the rock sugar orange are obtained using Hough transform. The K-means clustering pattern classification method is used to classify different circles, and the clustering results are corrected using a circle parameter correction model. © 2023 ACM.

关键词： Computer vision

Artificial Neural Microcircuits as Building Blocks: Concept and Challenges

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Walter, Andrew Wu, Shimeng Tyrrell, Andy M. McDaid, Liam McElholm, Malachy Sumithran, Nidhin Thandassery Harkin, Jim Trefzer, Martin A. School of Physics Engineering & Technology University of York United Kingdom School of Computing Engineering and Intelligent Systems Ulster University United Kingdom

Artificial Neural Networks (ANNs) are one of the most widely employed forms of bio-inspired computation. However the current trend is for ANNs to be structurally homogeneous. Furthermore, this structural homogeneity requires the application of complex training & learning tools that produce application specific ANNs, susceptible to pitfalls such as overfitting. In this paper, an new approach is explored, inspired by the role played in biology by Neural Microcircuits, the so called "fundamental processing elements" of organic nervous systems. How large neural networks, particularly Spiking Neural Networks (SNNs) can be assembled using Artificial Neural Microcircuits (ANMs), intended as off-the-shelf components, is articulated;the results of initial work to produce a catalogue of such Microcircuits though the use of Novelty Search is shown;followed by efforts to expand upon this initial work, including a discussion of challenges uncovered during these efforts & explorations of methods by which they might be overcome. Copyright © 2024, The Authors. All rights reserved.

关键词： Evolutionary algorithms

Evaluating Software Development Agents: Patch Patterns, Code Quality, and Issue Complexity in Real-World GitHub Scenarios

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Chen, Zhi Jiang, Lingxiao Centre for Research on Intelligent Software Engineering School of Computing and Information Systems Singapore Management University Singapore

In recent years, AI-based software engineering has progressed from pre-trained models to advanced agentic workflows, with Software Development Agents representing the next major leap. These agents, capable of reasoning, planning, and interacting with external environments, offer promising solutions to complex software engineering tasks. However, while much research has evaluated code generated by large language models (LLMs), comprehensive studies on agent-generated patches, particularly in real-world settings, are lacking. This study addresses that gap by evaluating 4,892 patches from 10 top-ranked agents on 500 real-world GitHub issues from SWE-Bench Verified, focusing on their impact on code quality. Our analysis shows no single agent dominated, with 170 issues unresolved, indicating room for improvement. Even for patches that passed unit tests and resolved issues, agents made different file and function modifications compared to the gold patches from repository developers, revealing limitations in the benchmark’s test case coverage. Most agents maintained code reliability and security, avoiding new bugs or vulnerabilities;while some agents increased code complexity, many reduced code duplication and minimized code smells. Finally, agents performed better on simpler codebases, suggesting that breaking complex tasks into smaller sub-tasks could improve effectiveness. This study provides the first comprehensive evaluation of agent-generated patches on real-world GitHub issues, offering insights to advance AI-driven software development. © 2024, CC BY.

关键词： Software design

Fingerspelling Classification for Robot Control 22

学校读者我要写书评

暂无评论

Fingerspelling Classification for Robot Control

22nd IEEE International Conference on Industrial Informatics, INDIN 2024

作者： McCready, Kevin Coleman, Sonya Kerr, Dermot Siddique, Nazmul Kerr, Emmett Aloimonos, Yiannis Fermüller, Cornelia School of Computing Engineering and Intelligent Systems Ulster University Londonderry United Kingdom Department of Electronic and Mechanical Engineering Atlantic Technological University Co.Donegal Letterkenny Ireland Institute for Advanced Computer Studies University of Maryland College ParkMD United States

ISBN: (纸本)9798331527471

Improvements to human-robot interaction methods could increase the ease of use of robots in manufacturing environments. Many of these environments are noisy and therefore preclude the use of audio communication between humans or in human-robot interactions. Therefore, this paper proposes using a gesture based communication system for robot control. To that end, the VGG16 and VGG19 convolutional neural network (CNN) structures are used for gesture classification along with 3 datasets of American Sign Language (ASL) fingerspelling images. The model performance is evaluated, and modifications made to their parameters to improve performance, before applying them to robot control tasks. The results show that with parameter tuning, test accuracies of up to, 100% are achievable. © 2024 IEEE.

关键词： Convolutional neural networks

Position and Attitude Control of Quadrotor UAV Based on a New Sliding Mode Control Strategy 42

学校读者我要写书评

暂无评论

Position and Attitude Control of Quadrotor UAV Based on a Ne...

42nd Chinese Control Conference, CCC 2023

作者： Wu, Yinghua Zhang, Ancai Pan, Guangyuan Du, Yingxue Liu, Zhi School of Automation and Electrical Engineering Linyi University Shandong Linyi276005 China Universities of Shandong Key Laboratory of Complex Systems and Intelligent Computing Linyi University Shandong Linyi276005 China

ISBN: (纸本)9789887581543

A Quadrotor unmanned aerial vehicle (UAV) is a typical underactuated nonlinear mechanical system with six degrees of freedom (DOF) and four inputs. This paper develops a new sliding mode control (SMC) strategy to solve the position and attitude control problem for this multi-DOF underactuated system. First, the dynamic motion equation model of the Quadrotor UAV is established. Second, the first and fourth inputs are designed by using the traditional SMC method and a novel Gaussian geometric function reaching law. And then, we design the second and third inputs based on a hierarchical SMC method. And the stability of sliding mode surfaces are analyzed using Lyapunov theory. The designed four sliding controllers guarantee the flight control objective of the Quadrotor UAV to be achieved. And the Gaussian reaching law used in the design of controllers can effectively reduce the chattering phenomena in the sliding mode surfaces. Numerical examples verify the effectiveness of our developed control strategy. © 2023 Technical Committee on Control Theory, Chinese Association of Automation.

关键词： Sliding mode control

Research on key technologies and application value of private cloud security cloud management platform 6

学校读者我要写书评

暂无评论

Research on key technologies and application value of privat...

6th International Conference on intelligent computing, Communication, and Devices, ICCD 2023

作者： Min, Xiao Mei, Guo School of Computer and Artificial Intelligence XiangNan University HuNan ChenZhou423000 China Advanced Embedded Computing and Intelligent Medical Systems Hunan Engineering Research Center HuNan ChenZhou423000 China

ISBN: (数字)9781510666269

ISBN: (纸本)9781510666252

Nowadays, with the rapid development and popularization of informatization and intelligence, "cloud computing service"is also born. Among them, "private cloud"is an important part of "cloud computing services". This is also known as the "internal cloud"or "corporate cloud"and plays an important role in the realization of information management within the enterprise. By studying the overall architecture of private cloud security management platform, this paper studies the key technologies of cloud security management platform and the application value of cloud security management platform. It can provide more flexible security solutions for governments and enterprises and provide better security for computer networks. © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

关键词： Cloud security