检索结果-内蒙古大学图书馆

FDAM: Filter-Dedicated Approximate Multiplier Design for Real-Time CNN Acceleration

IEEE Transactions on computer-Aided Design of Integrated Circuits and Systems 2025年

作者： Nguyen, Kihwan Kim Xuan Truong Jeon, Dongsuk Lee, Hyuk-Jae Korea Republic of program Department of Electrical Engineering Seoul08826 Korea Republic of Inter-University Semiconductor Research Center Seoul National University Graduate School of Convergence Science and Technology Research Institute for Convergence Science Seoul08826 Korea Republic of Department of Electrical and Computer Engineering Seoul08826 Korea Republic of

Video super-resolution (VSR) is widely used in various high-definition applications, such as HDTVs and smartphones, requiring a dedicated upscaling technique for realtime full-HD generation. To reduce on-chip buffers for largesize output feature maps, a streaming VSR accelerator may employ an output stationary dataflow, leading to a large energy consumption caused by frequent filter switching. To mitigate this, we introduce a new filter-dedicated multiplier design for realtime VSR acceleration. We replace costly multipliers with adders, shifters, and multiplexers (MUXes), referred to as unified multiple constant multiplications (UMCM). The conventional UMCM, however, may incur considerable area/power overhead due to the unified topology constraint among different filter sets. To address this problem, we propose a new approximated MCM (AMCM) problem to relax the constraint and an approximate compatible graph synthesis (A-CGS) framework to efficiently solve AMCM by jointly searching for approximated filters and constructing a unified graph. Additionally, we suggest a lightweight finetuning method by freezing approximated filters and only finetuning biases, which can recover the original model’s accuracy within a few epochs. Experimental results with synthetic data demonstrate that AMCM reduces the area by up to 49.8%, 44.8%, and 40.3% when considering constant sets of 2, 4, and 8, respectively. Our designs with SR application achieve up to a 73.3% reduction in energy consumption. Experiments with the Set5 and Set14 datasets show that our model with bias correction achieves similar restoration performance compared to the eightbit models. © 1982-2012 IEEE.

关键词： Acceleration

来源：评论

学校读者我要写书评

暂无评论

A Finite Difference Method for a Normalized Time-Fractional Black–Scholes Equation

引用

International Journal of Applied and Computational Mathematics 2025年第3期11卷 1-17页

作者： Nam, Yunjae Wang, Jian Lee, Chaeyoung Choi, Yongho Bang, Minjoon Li, Zhengang Kim, Junseok Program in Actuarial Science and Financial Engineering Korea University Seoul Republic of Korea School of Mathematics and Statistics Nanjing University of Information Science and Technology Nanjing China Department of Mathematics Kyonggi University Suwon Republic of Korea Computer and Information Engineering Daegu University Gyeongsan Republic of Korea Department of Mathematics Korea University Seoul Republic of Korea

In this article, we propose a normalized time-fractional Black–Scholes (TFBS) equation. The proposed model uses a normalized time-fractional derivative which has a distinctive feature wherein a weight function possesses the property that its integral with respect to time is always equal to one. This feature ensures a well-balanced integration over time and provides a fair comparison between different fractional orders. An implicit finite difference method is used for the numerical solution of the TFBS equation. We perform several standard numerical tests to examine the effect of the fractional parameter on the option pricing. These experiments are designed to investigate how variations in the fractional parameter influence pricing outcomes and provide valuable insights into the model’s behavior under different conditions. The computational results demonstrate the model’s potential for capturing more intricate market dynamics, which makes the proposed model a promising tool for financial analysis. In the appendix, we provide a computer program that implements the numerical methods discussed in the paper for interested readers.

关键词：

来源：评论

学校读者我要写书评

暂无评论

MERNet: Multimodal Emotion Recognition in Conversation Network via Cross-Modal Knowledge Distillation and Contrastive Learning

MERNet: Multimodal Emotion Recognition in Conversation Netwo...

引用

International Conference on Big Data and Smart Computing (BIGCOMP)

作者： Sooyeon Hwang Gahgene Gweon Eunseong Kwon Interdisciplinary Program in Artificial Intelligence Seoul National University South Korea Graduate School of Convergence Science and Technology Seoul National University South Korea Dept. of Computer Science and Engineering POSTECH South Korea

ISBN: (数字)9798331529024

ISBN: (纸本)9798331529031

Multimodal Emotion Recognition in Conversation (ERC) is a task of predicting the emotion of each utterance in a conversation by utilizing both verbal and non-verbal modalities. However, existing approaches often struggle to bridge cross-modal gaps, resulting in misaligned features and frequent misclassification of minority emotions into semantically similar majority emotions. To address these challenges, we propose MERNet, a framework that employs cross-modal knowledge distillation and contrastive learning to align multimodal features and effectively distinguish subtle emotions in conversations. Our framework consists of two stages: 1) guiding non-verbal modalities with the text modality to transfer knowledge and align their features, and 2) applying contrastive learning with emotion labels as anchors to distinguish subtle differences between similar emotions and address the class imbalance problem. Experiments conducted on two benchmark datasets, IEMOCAP and MELD, demonstrate that our MERNet outperforms existing state-of-the-art models.

关键词： Knowledge engineering Bridges Emotion recognition Visualization Computational modeling Contrastive learning Benchmark testing Linguistics Big Data

来源：评论

学校读者我要写书评

暂无评论

Alignment-Free Estimation of Read to Genome Distances and Its Applications

Alignment-Free Estimation of Read to Genome Distances and...

引用

29th International Conference on Research in Computational Molecular Biology, RECOMB 2025

作者： Şapcı, Ali Osman Berk Mirarab, Siavash Bioinformatics and Systems Biology Graduate Program UC San Diego San DiegoCA92093 United States Department of Electrical and Computer Engineering UC San Diego San DiegoCA92093 United States

ISBN: (纸本)9783031902512

Searching reads from unknown origins in a reference database and finding evolutionarily similar genomes is central to many applications. Quantifying the similarity by estimating the distance between each read and matching references could further help downstream analyses such as taxonomic characterization or even placing the read on a reference phylogeny. Such distances can be computed using alignment. Since alignment becomes impractical for ultra-large reference databases, many use k-mer-based tools for downstream tasks, but these methods do not compute distances. We introduce krepp to compute the distance between a read and a genome using a maximum likelihood framework built around k-mers. We further introduce algorithms to place reads on a reference tree. Empirical evaluations show that krepp accurately estimates distances, scales well with large databases, and can place short reads coming from any part of the genome on a reference phylogeny. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Evaluating the Systematic Reasoning Abilities of Large Language Models through Graph Coloring

arXiv

引用

arXiv 2025年

作者： Heyman, Alex Zylberberg, Joel Department of Electrical Engineering and Computer Science York University TorontoON Canada Jules Stein Eye Institute University of California Los AngelesCA United States Learning in Machines and Brains Program CIFAR TorontoON Canada

Contemporary large language models are powerful problem-solving tools, but they exhibit weaknesses in their reasoning abilities which ongoing research seeks to mitigate. We investigate graph coloring as a means of evaluating an LLM’s capacities for systematic step-by-step reasoning and possibility space exploration, as well as effects of semantic problem framing. We test Claude 3.5 Sonnet, Llama 3.1 405B, Gemini 1.5 Pro, GPT-4o, o1-mini, and DeepSeek-R1 on a dataset of k-coloring problems with 2 ≤ k ≤ 4 and vertex count 4 ≤ n ≤ 8, using partial algorithmic solvers to further categorize problems by difficulty. In addition to substantial but varying framing effects, we find that all models except o1-mini and R1 exhibit > 60% error rates on difficult problem types in all frames (> 15% for o1-mini and > 10% for R1), and no model achieves perfect accuracy even in the simple domain of 2-coloring 4-vertex graphs. Our results highlight both the considerable recent progress in LLM systematic reasoning and the limits of its reliability, especially in relation to increasing computational costs. We expect that more complex graph coloring problems, and procedural generation of arbitrary-complexity reasoning problems more broadly, offer further untapped potential for LLM benchmarking. © 2025, CC BY.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A novel biomechanical model of the proximal mouse forelimb predicts muscle activity in optimal control simulations of reaching movements

引用

Journal of Neurophysiology 2025年第4期133卷 1266-1278页

作者： Gilmer, Jesse I. Coltman, Susan K. Cuenu, Geraldine Hutchinson, John R. Huber, Daniel Person, Abigail L. Borno, Mazen Al Department of Computer Science and Engineering Computational Bioscience Program University of Colorado Denver j Anschutz Medical Campus Denver CO United States Department of Basic Neuroscience University of Geneva Geneva Switzerland Department of Physiology and Biophysics University of Colorado Denver j Anschutz Medical Campus Denver CO United States Department of Comparative Biomedical Sciences Royal Veterinary College London United Kingdom Department of Kinesiology The Pennsylvania State University University Park PA United States

Mice are key model organisms in neuroscience and motor systems physiology. Fine motor control tasks performed by mice have become widely used in assaying neural and biophysical motor system mechanisms. Although fine motor tasks provide useful insights into behaviors that require complex multi-joint motor control, there is no previously developed physiological biomechanical model of the adult mouse forelimb available for estimating kinematics, muscle activity, or kinetics during behaviors. Here, we developed a musculoskeletal model based on high-resolution imaging of the mouse forelimb that includes muscles spanning the neck, trunk, shoulder, and limbs. Physics-based optimal control simulations of the forelimb model were used to estimate in vivo muscle activity present when constrained to the tracked kinematics during reaching movements. The activity of a subset of muscles was recorded and used to assess the accuracy of the muscle patterning in simulation. We found that the synthesized muscle patterning in the forelimb model had a strong resemblance to empirical muscle patterning, suggesting that our model has utility in providing a realistic set of estimated muscle excitations over time when given a kinematic template. The strength of the similarity between empirical muscle activity and optimal control predictions increases as mice performance improves throughout learning of the reaching task. Our computational tools are available as open-source in the OpenSim physics and modeling platform. Our model can enhance research into limb control across broad research topics and can inform analyses of motor learning, muscle synergies, neural patterning, and behavioral research that would otherwise be inaccessible. NEW & NOTEWORTHY Investigations into motor planning and execution lack an accurate and complete model of the forelimb, which could bolster or expand on findings. We sought to construct such a model using high-detail scans of murine anatomy and prior research into

关键词： biophysical models motor control muscle patterning optimal control reaching behaviors

来源：评论

学校读者我要写书评

暂无评论

MathReader: Text-to-Speech for Mathematical Documents

arXiv

引用

arXiv 2025年

作者： Hyeon, Sieun Jung, Kyudan Kim, Nam-Joon Ryu, Hyun Gon Do, Jaeyoung Department of Electrical and Computer Engineering Seoul National University Korea Republic of Department of Mathematics Chung-Ang University Korea Republic of NVIDIA United States Interdisciplinary Program in Artificial Intelligence Seoul National University Korea Republic of

TTS (Text-to-Speech) document reader from Microsoft, Adobe, Apple, and OpenAI have been serviced worldwide. They provide relatively good TTS results for general plain text, but sometimes skip contents or provide unsatisfactory results for mathematical expressions. This is because most modern academic papers are written in LaTeX, and when LaTeX formulas are compiled, they are rendered as distinctive text forms within the document. However, traditional TTS document readers output only the text as it is recognized, without considering the mathematical meaning of the formulas. To address this issue, we propose MathReader, which effectively integrates OCR, a fine-tuned T5 model, and TTS. MathReader demonstrated a lower Word Error Rate (WER) than existing TTS document readers, such as Microsoft Edge and Adobe Acrobat, when processing documents containing mathematical formulas. MathReader reduced the WER from 0.510 to 0.281 compared to Microsoft Edge, and from 0.617 to 0.281 compared to Adobe Acrobat. This will significantly contribute to alleviating the inconvenience faced by users who want to listen to documents, especially those who are visually impaired. The code is available at https://***/hyeonsieun/MathReader. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

An End-to-End Framework for Neonatal Asphyxia Detection through Video Analysis

引用

IEEE Access 2025年 13卷 74689-74704页

作者： Kurnianingsih Nobukawa, Sou Widyawati, Melyana Nurul Pramana, Cipta Aji, Nurseno Bayu Thohari, Afandi Nur Aziz Hendrawati, Dwiana Sato-Shimokawara, Eri Kubota, Naoyuki Department of Electrical Engineering Politeknik Negeri Semarang H. Soedarto Tembalang Central Java Semarang50275 Indonesia Department of Computer Science Chiba Institute of Technology 2-17-1 Tsudanuma Chiba Narashino275-0016 Japan Department of Postgraduate Program Poltekkes Kemenkes Semarang Jl. Tirto Agung Pedalangan Kecamatan Banyumanik Central Java Semarang50268 Indonesia Department of Obstetrics and Gynecology KRMT Wongsonegoro Hospital Jl. Tembalang Central Java Semarang50272 Indonesia Department of Mechanical Engineering Politeknik Negeri Semarang H. Soedarto Tembalang Central Java Semarang50275 Indonesia Graduate School of Systems Design Tokyo Metropolitan University 6-6 Asahigaoka Hino Tokyo191-0065 Japan

Identifying asphyxia using computer vision in real-world settings poses challenges due to varying video quality, diverse lighting conditions, and subtle color changes in the newborn's skin. This study presents an end-to-end framework for automated neonatal asphyxia detection using time series video analysis and makes three key contributions. First, the proposed framework integrates YOLOv8-based instance segmentation with advanced feature extraction across multiple color spaces and texture analysis to detect neonatal asphyxia through the multi-modal analysis of skin features in video streams. Second, we introduce a new quality-aware temporal analysis framework that includes adaptive quality assessment for evaluating frames in real time, multi-stage feature stability tracking across temporal windows, hysteresis-based decision logic for ensuring temporal consistency, and LightGBM classification with comprehensive feature engineering to assess severity. Third, we provide a curated time series video dataset of 12,973 frames from 45 neonates, of which some were healthy, and some had asphyxia of varying severity. The findings show that the YOLOv8-based instance segmentation achieved a mean average precision (mAP@0.5) of 0.925 for accurate skin region isolation, and the LightGBM classifier outperformed traditional models with an accuracy of 0.998 and an F1-score of 0.998. The system maintains real-time processing at 30 FPS for normal and mild asphyxia cases with a minor reduction to 20 FPS in more challenging scenarios and exhibits robust temporal stability across severity levels, with consistency scores above 0.90. This framework has the potential to enhance neonatal care through continuous monitoring and timely intervention. © 2025 Institute of Electrical and Electronics Engineers Inc.. All rights reserved.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

AVTENet: A Human-Cognition-Inspired Audio-Visual Transformer-Based Ensemble Network for Video Deepfake Detection

引用

IEEE Transactions on Cognitive and Developmental Systems 2025年

作者： Hashmi, Ammarah Shahzad, Sahibzada Adil Lin, Chia-Wen Tsao, Yu Wang, Hsin-Min Taiwan International Graduate Program Academia Sinica Social Networks and Human-Centered Computing Program Taipei11529 Taiwan National Tsing Hua University Institute of Information Systems and Applications Hsinchu30013 Taiwan National Chengchi University Department of Computer Science Taipei11605 Taiwan National Tsing Hua University Department of Electrical Engineering Institute of Communications Engineering Hsinchu300044 Taiwan Academia Sinica Research Center for Information Technology Innovation Taipei11529 Taiwan

The recent proliferation of hyper-realistic deepfake videos has drawn attention to the threat of audio and visual forgeries. Most previous studies on detecting artificial intelligence-generated fake videos only utilize visual modality or audio modality. While some methods exploit audio and visual modalities to detect forged videos, they have not been comprehensively evaluated on multimodal datasets of deepfake videos involving acoustic and visual manipulations, and are mostly based on convolutional neural networks with low detection accuracy. Considering that human cognition instinctively integrates multisensory information including audio and visual cues to perceive and interpret content and the success of transformer in various fields, this study introduces the audio-visual transformer-based ensemble network (AVTENet). This innovative framework tackles the complexities of deepfake technology by integrating both acoustic and visual manipulations to enhance the accuracy of video forgery detection. Specifically, the proposed model integrates several purely transformer-based variants that capture video, audio, and audio-visual salient cues to reach a consensus in prediction. For evaluation, we use the recently released benchmark multimodal audio-video FakeAVCeleb dataset. For a detailed analysis, we evaluate AVTENet, its variants, and several existing methods on multiple test sets of the FakeAVCeleb dataset. Experimental results show that the proposed model outperforms all existing methods and achieves state-of-the-art performance on Testset-I and Testset-II of the FakeAVCeleb dataset. We also compare AVTENet against humans in detecting video forgery. The results show that AVTENet significantly outperforms humans. © 2016 IEEE.

关键词： audio-visual AVTENet Deepfake detection

来源：评论

学校读者我要写书评

暂无评论

Hybrid Deep Learning Model for Accurate Short-Term Electricity Price Forecasting

Hybrid Deep Learning Model for Accurate Short-Term Electrici...

引用

IEEE Texas Power and Energy Conference (TPEC)

作者： Arash Moradzadeh Mostafa Mouhammadpourfard Yang Weng Suhas Pol S. M. Muyeen Department of Electrical Engineering Qatar University Doha Qatar National Wind Institute Texas Tech University Lubbock TX USA School of Electrical Computer and Energy Engineering Arizona State University Tempe AZ USA Renewable Energy Program Texas Tech University Lubbock TX USA

ISBN: (数字)9798331541125

ISBN: (纸本)9798331541132

Accurate short-term electricity price forecasting (STEPF) is critical for efficient energy market operations, guiding investment strategies, resource allocation, and consumer behavior. This study introduces a hybrid deep learning approach specifically designed to improve STEPF accuracy by leveraging historical Hourly Ontario Energy Price (HOEP) data from 2017 to 2019. The model integrates advanced techniques, including data preprocessing and denoising through a Stacked Denoising Autoencoder (SDAE), along with enhanced temporal modeling via Bidirectional Long Short-Term Memory (BiLSTM) and Gated Recurrent Unit (GRU) networks. By capturing the complex dynamics inherent in electricity pricing data, the proposed hybrid model significantly enhances forecasting accuracy. Trained on data from 2017 and 2018, with 2019 used for testing, the model achieves a strong correlation coefficient (R = 99.86%) and substantially lowers forecasting errors. Comparative evaluations against established forecasting methods highlight the model's superior performance. This work demonstrates the practical value of deep learning techniques in the energy sector, particularly in responding to the volatility of demand and supply in real-time electricity markets.

关键词： Deep learning Accuracy Electricity Noise reduction Bidirectional long short term memory Predictive models Data models Stakeholders Forecasting Testing

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：