检索结果-内蒙古大学图书馆

Large receptive field strategy and important feature extraction strategy in 3D object detection

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Cui, Leichao Li, Xiuxian Meng, Min Jia, Guangyu Department of Control Science and Engineering College of Electronics and Information Engineering The Shanghai Research Institute for Intelligent Autonomous Systems Tongji University Shanghai201800 China The Centre for Robotics Research Department of Engineering King’s College London Strand LondonWC2R 2LS United Kingdom

The enhancement of 3D object detection is pivotal for precise environmental perception and improved task execution capabilities in autonomous driving. LiDAR point clouds, offering accurate depth information, serve as a crucial information for this purpose. Our study focuses on key challenges in 3D target detection. To tackle the challenge of expanding the receptive field of a 3D convolutional kernel, we introduce the Dynamic Feature Fusion Module (DFFM). This module achieves adaptive expansion of the 3D convolutional kernel’s receptive field, balancing the expansion with acceptable computational loads. This innovation reduces operations, expands the receptive field, and allows the model to dynamically adjust to different object requirements. Simultaneously, we identify redundant information in 3D features. Employing the Feature Selection Module (FSM) quantitatively evaluates and eliminates non-important features, achieving the separation of output box fitting and feature extraction. This innovation enables the detector to focus on critical features, resulting in model compression, reduced computational burden, and minimized candidate frame interference. Extensive experiments confirm that both DFFM and FSM not only enhance current benchmarks, particularly in small target detection, but also accelerate network performance. Importantly, these modules exhibit effective complementarity. Copyright © 2024, The Authors. All rights reserved.

关键词： Object detection

Are Structural Concepts Universal in Transformer Language Models? Towards Interpretable Cross-Lingual Generalization

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Xu, Ningyu Zhang, Qi Ye, Jingting Zhang, Menghan Huang, Xuanjing School of Computer Science Fudan University China Institute of Modern Languages and Linguistics Fudan University China Department of Chinese Language and Literature Fudan University China Research Institute of Intelligent Complex Systems Fudan University China

Large language models (LLMs) have exhibited considerable cross-lingual generalization abilities, whereby they implicitly transfer knowledge across languages. However, the transfer is not equally successful for all languages, especially for low-resource ones, which poses an ongoing challenge. It is unclear whether we have reached the limits of implicit cross-lingual generalization and if explicit knowledge transfer is viable. In this paper, we investigate the potential for explicitly aligning conceptual correspondence between languages to enhance cross-lingual generalization. Using the syntactic aspect of language as a testbed, our analyses of 43 languages reveal a high degree of alignability among the spaces of structural concepts within each language for both encoder-only and decoder-only LLMs. We then propose a meta-learning-based method to learn to align conceptual spaces of different languages, which facilitates zero-shot and few-shot generalization in concept classification and also offers insights into the cross-lingual in-context learning phenomenon. Experiments on syntactic analysis tasks show that our approach achieves competitive results with state-of-the-art methods and narrows the performance gap between languages, particularly benefiting those with limited resources. Copyright © 2023, The Authors. All rights reserved.

关键词： Syntactics

CROSS-ENTROPY IS ALL YOU NEED TO INVERT THE DATA GENERATING PROCESS

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Reizinger, Patrik Bizeul, Alice Juhos, Attila Vogt, Julia E. Balestriero, Randall Brendel, Wieland Klindt, David Max Planck Institute for Intelligent Systems Tübingen AI Center ELLIS Institute Tübingen Germany Department of Computer Science ETH Zürich ETH AI Center ETH Zürich Zürich Switzerland Department of Computer Science Brown University RI United States Cold Spring Harbor Laboratory Cold Spring Harbor New York United States

Supervised learning has become a cornerstone of modern machine learning, yet a comprehensive theory explaining its effectiveness remains elusive. Empirical phenomena, such as neural analogy-making and the linear representation hypothesis, suggest that supervised models can learn interpretable factors of variation in a linear fashion. Recent advances in self-supervised learning, particularly nonlinear Independent Component Analysis, have shown that these methods can recover latent structures by inverting the data generating process. We extend these identifiability results to parametric instance discrimination, then show how insights transfer to the ubiquitous setting of supervised learning with cross-entropy minimization. We prove that even in standard classification tasks, models learn representations of ground-truth factors of variation up to a linear transformation under a certain DGP. We corroborate our theoretical contribution with a series of empirical studies. First, using simulated data matching our theoretical assumptions, we demonstrate successful disentanglement of latent factors. Second, we show that on DisLib, a widely-used disentanglement benchmark, simple classification tasks recover latent structures up to linear transformations. Finally, we reveal that models trained on ImageNet encode representations that permit linear decoding of proxy factors of variation. Together, our theoretical findings and experiments offer a compelling explanation for recent observations of linear representations, such as superposition in neural networks. This work takes a significant step toward a cohesive theory that accounts for the unreasonable effectiveness of supervised learning. © 2024, CC BY.

关键词： Self-supervised learning

Adaptive RBF-Neural-Network Control with Force Observer for Teleoperation Robotic System 1st

学校读者我要写书评

暂无评论

Adaptive RBF-Neural-Network Control with Force Observer for...

1st International Conference on Cognitive Computation and systems, ICCCS 2022

作者： An, Zhaohui Min, Gaochen Jiang, Xiangzuo Yu, Xinbo He, Wei Silvestre, Carlos School of Intelligence Science and Technology Institute of Artificial Intelligence University of Science and Technology Beijing Beijing100083 China Beijing Advanced Innovation Center for Materials Genome Engineering University of Science and Technology Beijing Beijing100083 China Key Laboratory of Intelligent Bionic Unmanned Systems Ministry of Education University of Science and Technology Beijing Beijing100083 China Donald Bren School of Information and Computer Sciences University of California IrvineCA92697 United States Department of Electrical and Computer Engineering Faculty of Science and Technology University of Macau China LARSyS Instituto Superior Técnico Universidade de Lisboa Lisbon Portugal

ISBN: (纸本)9789819927883

In this paper, an adaptive radial basis function neural network (RBFNN) control strategy is proposed for bilateral teleoperation robotic system with uncertainty and time delay. Meantime, the force observer method is used to estimate the interaction force between the slave robot and environment. The model of teleoperation robotic system is analyzed, and then the uncertainty of bilateral teleoperation robotic system is estimated by using RBFNN. Using Lyapunov stability theorem, the stability of teleoperation robotic system under our proposed control is proved. Finally, the effectiveness of the proposed control algorithm is verified by Matlab Simulink, and position tracking and force estimation performance of the teleoperation robotic system are guaranteed. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Time delay

Decision-Making Framework For AMR Fleet Size In Manufacturing Environments

学校读者我要写书评

暂无评论

Decision-Making Framework For AMR Fleet Size In Manufacturin...

IEEE International Conference on Autonomous Robot systems and Competitions (ICARSC)

作者： Catarina Rema Romão Santos Henrique Piqueiro Diogo Miguel Matos Paulo Moura Oliveirat Pedro Costa Manuel F. Silva Fillipe L. Ribeiro CRIIS - Centre for Robotics in Industry and Intelligent Systems INESC TEC - Institute for Systems and Computer Engineering Technology and Science Porto Portugal Department of Engineering University of Trás-os-Montes and Alto Douro (UTAD) Vila Real Portugal Faculty of Engineering University of Porto (FEUP) Porto Portugal ISEP Polytechnic of Porto rua Dr. António Bernardino de Almeida Porto Portugal JPM Industry Vale de Cambra Portugal

ISBN: (数字)9798331538606

ISBN: (纸本)9798331538613

Industry 4.0 is transforming manufacturing environments, with robotics being a key technology that enhances various capabilities. The flexibility of Autonomous Mobile Robots has led to the rise of multi-robot systems in industrial settings. Considering the high cost of these robots, it is essential to determine the best fit of number and type before making any major investments. Simulation and modeling are valuable decision-support tools, allowing the simulation of different setups to address robot fleet sizing issues. This paper introduces a decision-support framework that combines a fleet manager software stack with the FlexSim simulator, helping decision-makers determine the most suitable mobile robots fleet size tailored to their needs. Unlike previous approaches, the developed solution integrates the same real robot coordination software in both simulation and actual deployment, ensuring that tested scenarios accurately reflect real-world conditions. A case study was conducted to evaluate the framework, involving multiple tasks of loading and unloading materials within a warehouse. Five different scenarios with varying fleet sizes were simulated, and their performances assessed. The analysis concluded that, for the case study under consideration, a fleet of three robots was the most suitable, considering relevant key performance indicators. The results confirmed that the developed solution is an effective alternative for addressing the problem and represents a novel technology with no prior state-of-the-art equivalents.

关键词： Service robots Robot kinematics Decision making Dynamic scheduling Software Batteries Software reliability Resource management Mobile robots Autonomous robots

An Optimized Ensemble Model for Prediction the Bandwidth of Metamaterial Antenna

学校读者我要写书评

暂无评论

computers, Materials & Continua 2022年第4期71卷 199-213页

作者： Abdelhameed Ibrahim Hattan F.Abutarboush Ali Wagdy Mohamed Mohamad Fouad El-Sayed M.El-kenawy Computer Engineering and Control Systems Department Faculty of EngineeringMansoura UniversityMansoura35516Egypt Electrical Engineering Department College of EngineeringTaibah UniversityMedina42353Saudi Arabia Operations Research Department Faculty of Graduate Studies for Statistical ResearchCairo UniversityGiza12613Egypt Wireless Intelligent Networks Center(WINC) School of Engineering and Applied SciencesNile UniversityGizaEgypt Department of Communications and Electronics Delta Higher Institute of Engineering and TechnologyMansoura35111Egypt Faculty of Artificial Intelligence Delta University for Science and TechnologyEgypt

Metamaterial Antenna is a special class of antennas that uses metamaterial to enhance their *** size affects the quality factor and the radiation loss of the *** antennas can overcome the limitation of bandwidth for small *** learning(ML)model is recently applied to predict antenna *** can be used as an alternative approach to the trial-and-error process of finding proper parameters of the simulated *** accuracy of the prediction depends mainly on the selected *** models combine two or more base models to produce a better-enhanced *** this paper,a weighted average ensemble model is proposed to predict the bandwidth of the Metamaterial *** base models are used namely:Multilayer Perceptron(MLP)and Support Vector Machines(SVM).To calculate the weights for each model,an optimization algorithm is used to find the optimal weights of the *** Group-Based Cooperative Optimizer(DGCO)is employed to search for optimal weight for the base *** proposed model is compared with three based models and the average ensemble *** results show that the proposed model is better than other models and can predict antenna bandwidth efficiently.

关键词： Metamaterial antenna machine learning ensemble model multilayer perceptron support vector machines

Position: Understanding LLMs Requires More Than Statistical Generalization

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Reizinger, Patrik Ujváry, Szilvia Mészáros, Anna Kerekes, Anna Brendel, Wieland Huszár, Ferenc Max Planck Institute for Intelligent Systems Tübingen Germany University of Cambridge United Kingdom AI Center UCL London United Kingdom Department of Computer Science ETH Zurich Switzerland Max Planck ETH Center for Learning Systems Switzerland ELLIS Institute Tübingen Germany Tübingen AI Center Germany

The last decade has seen blossoming research in deep learning theory attempting to answer, "Why does deep learning generalize?" A powerful shift in perspective precipitated this progress: the study of overparametrized models in the interpolation regime. In this paper, we argue that another perspective shift is due, since some of the desirable qualities of LLMs are not a consequence of good statistical generalization and require a separate theoretical explanation. Our core argument relies on the observation that AR probabilistic models are inherently non-identifiable: models zero or near-zero KL divergence apart—thus, equivalent test loss—can exhibit markedly different behaviors. We support our position with mathematical examples and empirical observations, illustrating why non-identifiability has practical relevance through three case studies: (1) the non-identifiability of zero-shot rule extrapolation;(2) the approximate non-identifiability of in-context learning;and (3) the non-identifiability of finetunability. We review promising research directions focusing on LLM-relevant generalization measures, transferability, and inductive biases. Copyright © 2024, The Authors. All rights reserved.

关键词： Deep learning

An Empirically Grounded Identifiability Theory Will Accelerate Self-Supervised Learning Research

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Reizinger, Patrik Balestriero, Randall Klindt, David Brendel, Wieland Max Planck Institute for Intelligent Systems Tübingen Germany ELLIS Institute Tübingen Germany Department of Computer Science Brown University RI United States Cold Spring Harbor Laboratory Cold Spring HarborNY United States Tübingen AI Center Germany

Self-Supervised Learning (SSL) powers many current AI systems. As research interest and investment grow, the SSL design space continues to expand. The Platonic view of SSL, following the Platonic Representation Hypothesis (PRH), suggests that despite different methods and engineering approaches, all representations converge to the same Platonic ideal. However, this phenomenon lacks precise theoretical explanation. By synthesizing evidence from Identifiability Theory (IT), we show that the PRH can emerge in SSL. However, current IT cannot explain SSL’s empirical success. To bridge the gap between theory and practice, we propose expanding IT into what we term Singular Identifiability Theory (SITh), a broader theoretical framework encompassing the entire SSL pipeline. SITh would allow deeper insights into the implicit data assumptions in SSL and advance the field towards learning more interpretable and generalizable representations. We highlight three critical directions for future research: 1) training dynamics and convergence properties of SSL;2) the impact of finite samples, batch size, and data diversity;and 3) the role of inductive biases in architecture, augmentations, initialization schemes, and optimizers. © 2025, CC BY.

关键词： Engineering research

Large-Scale Radiomics Analysis for Prostate Cancer Detection Harnessing Machine and Deep Learning Models

学校读者我要写书评

暂无评论

Large-Scale Radiomics Analysis for Prostate Cancer Detection...

IEEE International Symposium on Biomedical Imaging

作者： Eugenia Mylona Dimitrios I. Zaridis Charalampos N. Kalantzopoulos Vasileios C. Pezoulas Nikolaos S. Tachos George K. Matsopoulos Daniele Regge Nikolaos Papanikolaou Manolis Tsiknakis Kostas Marias Dimitrios I. Fotiadis Biomedical Research Institute FORTH Ioannina Greece Unit of Medical Technology Intelligent Information Systems University of Ioannina Greece Biomedical Engineering Laboratory School of Electrical & Computer Engineering National Technical University of Athens Greece Institute of Computer Science FORTH Heraklion Greece Department of Radiology Candiolo Cancer Institute FPO-IRCCS Strada Provinciale Candiolo Italy Computational Clinical Imaging Group Champalimaud Foundation Portugal Computational Biomedicine Laboratory Institute of Computer Science FORTH Heraklion Greece Department of Electrical and Computer Engineering Hellenic Mediterranean University Heraklion Greece

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

Prostate cancer (PCa) is the second most common cancer in men, making early detection crucial for effective treatment. This study explores the efficacy of machine learning (ML) and deep learning (DL) models in detecting PCa lesions using radiomics features derived from multiparametric MRI data. Leveraging a dataset of 7621 retrospective and 2132 prospective cases from 12 clinical centers across Europe, we compared the performance of various ML (LASSO, Random Forest, SVM, XGBoost) and DL models (CNN, MLP, Transformer, CNN&MLP). Radiomic features were extracted and refined through a robust feature selection process. Models' performance was evaluated through nested cross-validation (CV) and externally. The CNN&MLP model achieved the highest ROC AUC (0.86) and PR AUC (0.90) in CV, while LASSO showed consistent generalization in external validation. Our findings indicate that DL models excel in high-dimensional data analysis but face challenges in generalizing to external datasets, highlighting the importance of external validation for clinical applicability.

关键词： Deep learning Support vector machines Analytical models Biological system modeling Feature extraction Transformers Lesions Prostate cancer Radiomics Principal component analysis