检索结果-内蒙古大学图书馆

Low-temperature spin Seebeck effect in nonmagnetic vanadium dioxide

Physical Review B 2024年第2期110卷 024415-024415页

作者： Renjie Luo Tanner J. Legvold Liyang Chen Henry Navarro Ali C. Basaran Deshun Hong Changjiang Liu Anand Bhattacharya Ivan K. Schuller Douglas Natelson Department of Physics and Astronomy Applied Physics Graduate Program Rice Quantum Institute Department of Physics and Center for Advanced Nanoscience Materials Science Division Department of Electrical and Computer Engineering Department of Materials Science and NanoEngineering

The spin Seebeck effect (SSE) is sensitive to thermally driven magnetic excitations in magnetic insulators. Vanadium dioxide in its insulating low-temperature phase is expected to lack magnetic degrees of freedom, as vanadium atoms are thought to form singlets upon dimerization of the vanadium chains. Instead, we find a paramagnetic SSE response in VO2 films that grows as the temperature decreases below 50 K. The field and temperature-dependent SSE voltage is qualitatively consistent with a general model of paramagnetic SSE response and inconsistent with triplet spin transport. Quantitative estimates find a spin Seebeck coefficient comparable in magnitude to that observed in strongly magnetic materials. The microscopic nature of the magnetic excitations in VO2 requires further examination.

关键词： Spin Seebeck effect Paramagnets Strongly correlated systems

来源：评论

学校读者我要写书评

暂无评论

Seafloor Imaging Maps by Incorporating Side Scan Sonar Technology on Hybrid Autonomous Underwater Glider 6

Seafloor Imaging Maps by Incorporating Side Scan Sonar Techn...

引用

6th International Seminar on Research of Information Technology and Intelligent Systems, ISRITI 2023

作者： Atmaja, Ardian Budi Kusuma Hidayat, Egi Muhammad Idris Riyanto Trilaksono, Bambang Siregar, Simon Habibullah, Natsir School of Electrical Engineering and Informatics Institut Teknologi Bandung Bandung Indonesia Institut Teknologi Bandung Telkom University Doctoral Program of Electrical Engineering Department Computer Engineering Faculty of Applied Science Indonesia

ISBN: (纸本)9798350358346

This research aims to construct a two-dimensional image to represent an underwater geometry map with a Side Scan Sonar (SSS) mounted on a Hybrid Autonomous Underwater Glider (HAUG). Building the underwater map has two stages of the process. The first stage is preprocessing which includes Time-varying gains (TVG), slant range correction, and ground range correction. The second is HAUG navigation, the movement of HAUG orientation to the angle and distance of SSS readings and plotting the geometry of the grid map. With these two stages, a two-dimensional grid map will be formed. Meanwhile, the 3D plot utilizes slant range and ground range correction to obtain the surface height of the bottom of the pool. From the experiments conducted to conduct mapping in the pool, two-dimensional and three-dimensional visual forms of the pool floor and pool walls were produced. The results of the estimated measurement of the width of the pool obtained 14.05 meters and the height of the pool 3.199 meters with an SSS beam angle of 58 degrees. © 2023 IEEE.

关键词： Geometry

来源：评论

学校读者我要写书评

暂无评论

The Indoor People Tracking and Counting System

The Indoor People Tracking and Counting System

引用

2023 International Automatic Control Conference, CACS 2023

作者： Lin, Hao-Pu Han, Chin-Chuan Liao, Jun-Wei Ciou, Chong-Han Hong, Yi-Cheng Tan, Ming-Lun National United University Ph. D. Program in Material and Chemical Engineering MiaoLi Taiwan National United University Department of Computer Science and Information Engineering MiaoLi Taiwan

ISBN: (纸本)9798350306354

COVID-19 outbreaks and becomes serious from 2019 winter. Taking body temperature, wearing masks, recording footprint and avoiding crowds become important tasks and require a lot of manpower support for the prevention of the epidemic in every. In this study, we have integrated the infrared thermometer devices to measure people temperature and the people identities through RGB images at the entrance of buildings. Using the computer vision and deep learning methods, we would like to build a 24/7 and region-wide coverage assistance system to automatically record the footprint of a specific people in the building. We also detect if people taking off their masks in the yard at any time. The processing videos are grabbed from the existing CCTV systems. The developed system also locates the crowd and calculate the number of people in a specified area, and the manager can know the number of people and control the access in real time. It will reduce the burden of manpower. In addition, we have optimized the system performance to handle multiple camera videos and to reduce the hardware cost at the same time. © 2023 IEEE.

关键词： Access control

来源：评论

学校读者我要写书评

暂无评论

Flexible mapping of abstract domains by grid cells via self-supervised extraction and projection of generalized velocity signals 38

Flexible mapping of abstract domains by grid cells via self-...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Iyer, Abhiram Chandra, Sarthak Sharma, Sugandha Fiete, Ila Department of Electrical Engineering and Computer Science MIT CambridgeMA United States Department of Brain and Cognitive Sciences MIT CambridgeMA United States McGovern Institute for Brain Research MIT CambridgeMA United States MIT CambridgeMA United States

Grid cells in the medial entorhinal cortex create remarkable periodic maps of explored space during navigation. Recent studies show that they form similar maps of abstract cognitive spaces. Examples of such abstract environments include auditory tone sequences in which the pitch is continuously varied or images in which abstract features are continuously deformed (e.g., a cartoon bird whose legs stretch and shrink). Here, we hypothesize that the brain generalizes how it maps spatial domains to mapping abstract spaces. To sidestep the computational cost of learning representations for each high-dimensional sensory input, the brain extracts self-consistent, low-dimensional descriptions of displacements across abstract spaces, leveraging the spatial velocity integration of grid cells to efficiently build maps of different domains. Our neural network model for abstract velocity extraction factorizes the content of these abstract domains from displacements within the domains to generate content-independent and self-consistent, low-dimensional velocity estimates. Crucially, it uses a self-supervised geometric consistency constraint that requires displacements along closed loop trajectories to sum to zero, an integration that is itself performed by the downstream grid cell circuit over learning. This process results in high fidelity estimates of velocities and allowed transitions in abstract domains, a crucial prerequisite for efficient map generation in these high-dimensional environments. We also show how our method outperforms traditional dimensionality reduction and deep-learning based motion extraction networks on the same set of tasks. This is the first neural network model to explain how grid cells can flexibly represent different abstract spaces and makes the novel prediction that they should do so while maintaining their population correlation and manifold structure across domains. Fundamentally, our model sheds light on the mechanistic origins of cognitive flexib

关键词：

来源：评论

学校读者我要写书评

暂无评论

Head-wise Shareable Attention for Large Language Models

Head-wise Shareable Attention for Large Language Models

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Cao, Zouying Yang, Yifei Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University China Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University China Shanghai Key Laboratory of Trusted Data Circulation and Governance in Web3 China

ISBN: (纸本)9798891761681

Large Language Models (LLMs) suffer from huge number of parameters, which restricts their deployment on edge devices. Weight sharing is one promising solution that encourages weight reuse, effectively reducing memory usage with less performance drop. However, current weight sharing techniques primarily focus on small-scale models like BERT and employ coarse-grained sharing rules, e.g., layer-wise. This becomes limiting given the prevalence of LLMs and sharing an entire layer or block obviously diminishes the flexibility of weight sharing. In this paper, we present a perspective on head-wise shareable attention for large language models. We further propose two memory-efficient methods that share parameters across attention heads, with a specific focus on LLMs. Both of them use the same dynamic strategy to select the shared weight matrices. The first method directly reuses the pre-trained weights without retraining, denoted as DirectShare. The second method first post-trains with constraint on weight matrix similarity and then shares, denoted as PostShare. Experimental results reveal our head-wise shared models still maintain satisfactory capabilities, demonstrating the feasibility of fine-grained weight sharing applied to LLMs. © 2024 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

A Pilot Study on Clinician-AI Collaboration in Diagnosing Depression from Speech

arXiv

引用

arXiv 2024年

作者： Feng, Kexin Chaspari, Theodora Department of Computer Science and Engineering Texas A&M University College StationTX United States Institute of Cognitive Science Department of Computer Science University of Colorado Boulder BoulderCO United States

This study investigates clinicians’ perceptions and attitudes toward an assistive artificial intelligence (AI) system that employs a speech-based explainable ML algorithm for detecting depression. The AI system detects depression from vowel-based spectrotemporal variations of speech and generates explanations through explainable AI (XAI) methods. It further provides decisions and explanations at various temporal granularities, including utterance groups, individual utterances, and within each utterance. A small-scale user study was conducted to evaluate users’ perceived usability of the system, trust in the system, and perceptions of design factors associated with several elements of the system. Quantitative and qualitative analysis of the collected data indicates both positive and negative aspects that influence clinicians’ perception toward the AI. Results from quantitative analysis indicate that providing more AI explanations enhances user trust but also increases system complexity. Qualitative analysis indicates the potential of integrating such systems into the current diagnostic and screening workflow, but also highlights existing limitations including clinicians’ reduced familiarity with AI/ML systems and the need for user-friendly and intuitive visualizations of speech information. © 2024, CC BY.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Word formation supports efficient communication: The case of compounds 44

Word formation supports efficient communication: The case of...

引用

44th Annual Meeting of the cognitive science Society: cognitive Diversity, CogSci 2022

作者： Xu, Aotao Kemp, Charles Frermann, Lea Xu, Yang Department of Computer Science University of Toronto Canada School of Computing and Information Systems University of Melbourne Australia School of Psychological Sciences University of Melbourne Australia Cognitive Science Program University of Toronto Canada

Compounding is a common type of word formation extensively studied in linguistics and cognitive psychology. A growing line of research suggests that the lexicon supports efficient communication by balancing informativeness and simplicity. We propose that the formation of novel compounds reflects a similar tradeoff between informativeness and word length. We formalize this hypothesis in information-theoretic terms and develop a computational procedure to evaluate our hypothesis on English noun compounds that emerged over the past century. We find that attested compounds achieve more efficient tradeoffs between informativeness and word length than do alternative word forms. Our work demonstrates how word formation and compositionality can be connected with information-theoretic approaches to the design of the lexicon. © 2022 The Author(s). This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY)

关键词： Information theory

来源：评论

学校读者我要写书评

暂无评论

LaCo: Large Language Model Pruning via Layer Collapse

LaCo: Large Language Model Pruning via Layer Collapse

引用

2024 Conference on Empirical Methods in Natural Language Processing, EMNLP 2024

作者： Yang, Yifei Cao, Zouying Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University China Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University China Shanghai Key Laboratory of Trusted Data Circulation and Governance in Web3 China

ISBN: (纸本)9798891761681

Large language models (LLMs) based on transformer are witnessing a notable trend of size expansion, which brings considerable costs to both model training and inference. However, existing methods such as model quantization, knowledge distillation, and model pruning are constrained by various issues, including hardware support limitations, the need for extensive training, and alterations to the model internal structure. In this paper, we propose a concise layer-wise structured pruner called Layer Collapse (LaCo), in which rear model layers collapse into a prior layer, enabling a rapid reduction in model size while preserving the model structure. Comprehensive experiments show that our method maintains an average task performance of over 80% at pruning ratios of 25-30%, significantly outperforming existing state-of-the-art structured pruning methods. We also conduct post-training experiments to confirm that the LaCo effectively inherits the parameters of the original model. Additionally, we perform ablation studies on various settings of LaCo. Finally, we discuss our motivation from the perspective of layer-wise similarity and evaluate the performance of the pruned LLMs across various pruning ratios. © 2024 Association for Computational Linguistics.

关键词： Computational linguistics

来源：评论

学校读者我要写书评

暂无评论

Remote Sensing Scene Classification Based on Hybrid Convolutional Neural Network and Attention Module

Remote Sensing Scene Classification Based on Hybrid Convolut...

引用

Advancement in Data science, E-learning and Information Systems (ICADEIS), International Conference on

作者： Muhammad Yusuf Kardawi Alexander Agung Santoso Gunawan Profesional Engineer Program Faculty of Engineering Bina Nusantara University Jakarta Indonesia Computer Science Department Faculty of Computer Science University of Indonesia Jakarta Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia

ISBN: (数字)9798331513320

ISBN: (纸本)9798331513337

Remote sensing images present classification challenges due to the complexity of their structural and spatial patterns. This research explores a hybrid approach that combines convolutional neural network (CNN) and attention through feature fusion to improve scene classification accuracy in remote sensing images. The proposed architecture utilizes EfficientNet and VGGNet to extract depth features separately. The extracted features are then integrated with Dynamic Selfattention (DSA), which dynamically focuses the model on the most relevant information in the image. DSA allows the model to adaptively assign weights to different parts of the image, thus improving the discriminative ability of the model. Furthermore, a feature fusion technique is applied to combine information from different layers of the CNN and DSA modules. Experiments conducted on the UC Merced dataset showed accuracies of 0.9181 and 0.9167. These results show that the combination of CNN, DSA, and feature fusion is an effective and robust approach for remote sensing image classification.

关键词： Adaptation models Scene classification Accuracy Electronic learning Data science Feature extraction Convolutional neural networks Data mining Remote sensing Information systems

来源：评论

学校读者我要写书评

暂无评论

A Fitness Movement Evaluation System Using Deep Learning 15

A Fitness Movement Evaluation System Using Deep Learning

引用

15th International Conference on Knowledge and Systems engineering, KSE 2023

作者： Chang, Chin-Chih Wei, Chi-Hung Wu, Hao-Wei Hsiao, Sean Chung Hua University Department of Computer Science and Information Engineering Hsinchu City Taiwan Ph.D. Program in Engineering Science Chung Hua University Hsinchu City Taiwan Bachelor Program in Applied Foreign Language and Smart Commerce Chung Hua University Hsinchu City Taiwan

ISBN: (纸本)9798350329742

This paper proposes a fitness movement evaluation system using deep learning. The system uses a deep convolutional neural network (CNN) to extract features from pictures of fitness movements. The features are then used to classify the movements into different categories. The system is evaluated on a dataset of pictures of fitness movements. The results show that the system can accurately classify the movements into different categories. The system is designed to provide feedback to users on their fitness movements. The proposed system is a valuable tool for fitness enthusiasts. The main contribution of this paper is to propose a way to give users a score for their fitness movement. It can help users improve their fitness and track their progress over time. The system is also a valuable tool for fitness professionals. It can help professionals develop new fitness programs and provide feedback to their clients. © 2023 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：