检索结果-内蒙古大学图书馆

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Liu, Lanxin Liu, Xinyan Li, Guorong School of Computer Science and Technology Key Lab of Big Data Mining and Knowledge Management University of Chinese Academy of Sciences China School of Computer Science and Technology Harbin Institute of Technology Weihai China

ISBN: (纸本)9798350368741

Few-shot object counting and detection aim to count objects along with their bounding boxes specified by exemplar bounding boxes. Current mainstream methods predict density maps by applying similarity between exemplar and image features to get counting results and detecting peak points from the density map as the positions of objects. However, the sub-parts of objects can also have high similarity to the exemplars, harming the counting and detection performance. To address these issues, we propose a two-stage Subpart Suppression Network (SSN), consisting of a Subpart Suppression Density map Predictor (SSDP), which enforces the model focus on whole objects rather than subparts, and a SAM-based Detector and Verifier, which refines the final prediction by a clustering method. Extensive experiments on two popular datasets show an advantage in performance over the state-of-the-art methods and prove the components' effectiveness. © 2025 IEEE.

关键词： Counting Detection Few-Shot Localization

来源：评论

学校读者我要写书评

暂无评论

A Method of Enhancing Heterogeneous Graph Representation for Predicting the Associations between lncRNAs and Diseases

引用

Current Bioinformatics 2025年

作者： Yao, Dengju Wu, Yuehu Zhan, Xiaojuan School of Computer Science and Technology Harbin University of Science and Technology Harbin China College of Computer Science and Technology Heilongjiang Institute of Technology Harbin China

Background: Long non-coding RNAs (lncRNAs) are a category of more extended RNA strands that lack protein-coding abilities. Although they are not involved in the translation of proteins, studies have shown that they play essential regulatory functions in cells, regulating gene expression and cell biological processes. However, it is both costly and inefficient to determine the associations between lncRNAs and diseases through biological experiments. Therefore, there is an urgent need to develop convenient and fast computational methods to predict lncRNA-disease associations (LDAs) more efficiently. Objective: Predicting disease-associated lncRNAs can help explore the mechanisms of action of lncRNAs in diseases, and this is crucial for early intervention and treatment of diseases. Methods: In this paper, we propose an enhanced heterogeneous graph representation method for predicting LDAs, named GCGALDA. The GCGALDA first obtains the topological structure features of nodes by a biased random walk. Based on this, the neighboring nodes of a node are weighted using the attention mechanism to further mine the semantic association relationships between nodes in the graph data. Then, a graph convolution network (GCN) is used to transfer the neighborhood features of the node to the central node and combine them with the node's features so that the final node representation contains not only structural information but also semantic association information. Finally, the association score between lncRNA and disease is obtained by multilayer perceptron (MLP). Results: As evidenced by the experimental findings, the GCGALDA outperforms other advanced models in terms of prediction accuracy on openly accessible databases. In addition, case studies on several human diseases further confirm the predictive ability of the GCGALDA. Conclusion: In conclusion, the proposed GCGALDA model extracts multi-perspective features, such as topology, semantic association, and node attributes, obtains

关键词： attention mechanism biased random walk graph convolution Heterogeneous graph semantic association structural information topological structure features

来源：评论

学校读者我要写书评

暂无评论

Hyperspectral Image Cross Scene Classification Combined Domain Adaptation with XGBoost

Hyperspectral Image Cross Scene Classification Combined Doma...

引用

2024 International Conference on Measurement, Communication, and Virtual Reality, MCVR 2024

作者： Liu, Chengyang Wu, Haibin Li, Jiarong Liu, Xinyu Yan, Siqi Wang, Aili Heilongjiang Province Key Laboratory of Laser Spectroscopy Technology and Application College of Measurement and Control Technology and Communication Engineering Harbin University of Science and Technology Harbin150080 China School of Electronics and Information Engineering Harbin Institute of Technology Harbin150001 China

ISBN: (纸本)9781510691773

To address the issues of domain discrepancies and limited labeled samples in hyperspectral image classification of cross scene, this paper proposes an ensemble learning framework that integrates eXtreme Gradient Boosting (XGBoost) with confidence domain adaptation strategies to improve classification performance. First, the Houston dataset from different scenes is input separately into a deep embedding model. The hyperspectral images are embedded into a low-dimensional space, ensuring that similar spectra are represented more closely in this space, thereby capturing the intrinsic structure and distribution of the hyperspectral image dataset. Subsequently, these samples are mapped into a discriminator model, where domain adaptation techniques are employed to reduce differences in feature distributions across different scenes. Additionally, confidence-based learning is used to identify and remove noisy samples in the Houston dataset, thereby enhancing the alignment of feature distributions among various scenes. Finally, through ensemble learning, a weighted voting strategy is employed to combine multiple base classifiers into a strong classification model. This approach better captures the complex features within the data, thereby reducing the bias of individual classifiers and enhancing the overall generalization ability and robustness of the model. Experimental results on the Houston dataset demonstrate that the proposed method outperforms popular domain adaptation learning models and traditional methods in terms of classification accuracy and robustness, showcasing its potential and advantages in practical applications. This result confirms the effectiveness of combining XGBoost and confidence domain adaptation strategies in handling hyperspectral image classification. © 2025 SPIE.

关键词： Photointerpretation

来源：评论

学校读者我要写书评

暂无评论

Interference-Based Reliability and Capacity Analysis for IEEE 802.11 Broadcast Ad-Hoc Networks on the Highway 11

Interference-Based Reliability and Capacity Analysis for IEE...

引用

11th International Conference on Vehicle technology and Intelligent Transport Systems, VEHITS 2025

作者： Li, Zhijuan Wu, Xintong Li, Xiaokun Ma, Xiaomin Department of Computer and Big Data Heilongjiang University Harbin150080 China Postdoctoral Program of Heilongjiang Hengxun Technology Co. Ltd. Harbin150090 China School of Computer Science and Technology Harbin Institute of Technology Harbin150001 China College of Science and Engineering Oral Roberts University TulsaOK74171 United States

ISBN: (纸本)9789897587450

Interference is a critical factor that degrades wireless network performance. In IEEE 802.11 wireless broadcast networks, hidden terminals and concurrent transmissions are the primary sources of interference due to the carrier sense multiple access (CSMA) mechanism. Therefore, analyzing the signal-to-interference-plus-noise ratio (SINR) distribution is essential for evaluating network performance, whose derivation relates to the transmission probabilities of hidden terminals and concurrent transmissions. In this study, we utilize the existing semi-markov process (SMP) model to calculate these transmission probabilities. Subsequently, we employ the Laplace transform to analyze the SINR distribution in IEEE 802.11 broadcast ad-hoc networks on highways. Based on the derived SINR distribution, we further evaluate the reliability and capacity. This approach can be readily extended to two-dimensional (2D) or three-dimensional (3D) scenarios by employing d-dimensional (1 ≤ d ≤ 3, dD) point process. Experimental results demonstrate that the proposed model achieves high accuracy under small to medium interference ranges. Additionally, the analysis remains highly accurate for receivers within 70 meters, even in scenarios with large interference ranges. Copyright © 2025 by SCITEPRESS - science and technology Publications, Lda.

关键词： Markov processes

来源：评论

学校读者我要写书评

暂无评论

New Cooling Structure for Improving the Oil Cooling Effects of the End Winding of Flat Winding 19th

New Cooling Structure for Improving the Oil Cooling Effects ...

引用

19th Annual Conference of China Electrotechnical Society, ACCES 2024

作者： Cai, Wei Li, Guicang Yang, Yongxi Gong, Haigui School of Electrical Engineering Harbin University of Science and Technology Harbin150080 China School of Vehicle and Mechanical Engineering Beijing Institute of Technology Beijing100081 China

ISBN: (纸本)9789819613786

The oil-cooled flat-wire winding machines characterized with high electromagnetic loads and high power densities are widely used in for electric vehicle. However, high AC losses and large differences between the interlayer are inevitable when the motor operated in high-speed operation, posing serious challenges to reliable cooling of the motors. The transported winding with multi-segmented winding could significantly suppress the AC loss, but its end winding space structure is complex and might affect the cooling effect of winding. In this paper, the electrical machine with end transported winding is taken as the research object. The fluid-solid coupling model with oil and gas two-phase flow for the end transported winding is established, and the fluid distribution and local hot spot distribution are then analyzed. A new oil-injection structure is then proposed and are optimized with the Taguchi approach. The simulation result shows that the maximum temperature of the end winding is reduced by approximately 20 ℃ with the proposed oil-injection structure. © Beijing Paike Culture Commu. Co., Ltd. 2025.

关键词： Machine windings

来源：评论

学校读者我要写书评

暂无评论

Incorporate Global Information from Entire Datasets for Knowledge Tracing via Mini-Batch Input

Incorporate Global Information from Entire Datasets for Know...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Zhao, Hui Fu, Tingyu Wangxuan Institute of Computer Technology Peking University Beijing China School of Light Industry Harbin University of Commerce Harbin China

ISBN: (纸本)9798350368741

Knowledge tracing uses students' answer record data and the relationship between exercises to predict students' future answering performance. However, in the deep learning model, the input manner of mini-batch may prevent the network from learning global information such as the difficulty of exercises. In this paper, we propose a method to add global information into mini-batches. The added global information is obtained from the entire dataset by using a specific algorithm. Experimental results show that on three classic datasets, the model performance is improved by about 3.9% (average) on the AUC scale. Additionally, the artificially designed global information enhances the interpretability of the model. © 2025 IEEE.

关键词： global information knowledge tracing representation learning

来源：评论

学校读者我要写书评

暂无评论

LoRA-drop: Efficient LoRA Parameter Pruning based on Output Evaluation 31

LoRA-drop: Efficient LoRA Parameter Pruning based on Output ...

引用

31st International Conference on Computational Linguistics, COLING 2025

作者： Zhou, Hongyun Lu, Xiangyu Xu, Wang Zhu, Conghui Zhao, Tiejun Yang, Muyun Faculty of Computing Harbin Institute of Technology China Department of Computer Science & Technology Tsinghua University Beijing China

ISBN: (纸本)9798891761964

Low-Rank Adaptation (LoRA) is currently the most commonly used Parameter-efficient fine-tuning (PEFT) method. However, it still faces high computational and storage costs to models with billions of parameters. Most previous studies have tackled this issue by using pruning techniques. Nonetheless, these efforts only analyze LoRA parameter features to evaluate their importance, such as parameter count, size, and gradient. In fact, the output of LoRA directly impacts the fine-tuned model. Preliminary experiments indicate that a fraction of LoRA possesses significantly high output values, substantially influencing the layer output. Motivated by the observation, we propose LoRA-drop. Concretely, LoRA-drop evaluates the importance of LoRA based on the LoRA output. Then we retain LoRA for important layers and the other layers share the same LoRA. We conduct abundant experiments with models of different scales on NLU and NLG tasks. Results demonstrate that LoRA-drop can achieve performance comparable to full fine-tuning and LoRA while retaining 50% of the LoRA parameters on average. © 2025 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

LightFinder: Finding Persistent Items with Small Memory 20th

LightFinder: Finding Persistent Items with Small Memory

引用

20th IFIP WG 10.3 International Conference on Network and Parallel Computing, NPC 2024

作者： Cao, Lu Xiao, Weiqiang Zhang, Weizhe Department of Computer Science and Technology Harbin Institute of Technology Shenzhen China Peng Cheng Laboratory Shenzhen China

ISBN: (纸本)9789819628292

Detecting persistent items in large-scale data streams efficiently and accurately is a significant challenge, particularly when working with limited memory. Current state-of-the-art methods often require substantial memory resources to maintain high detection accuracy. To address this limitation, we present LightFinder (LF), which leverages a Conservative-Update (CU) strategy to minimize hash collision errors by selectively updating the smallest active counter. Additionally, LF supports detection across varying persistence windows, enhancing detection accuracy and flexibility in skewed data streams. Our analysis reveals that LF excels in detecting persistent items with high accuracy even under stringent memory constraints. Our approach effectively estimates item persistence and maintains robust performance across various persistence windows. Notably, when operating with just 10% of the memory required by On-Off Sketch (OO), LF achieves an Average Absolute Error (AAE) nearly half that of OO (up to 1.5 times lower), and its Average Relative Error (ARE) is reduced by an order of magnitude. This demonstrates LF significantly outperforms existing methods in terms of space efficiency. Our codes are open-sourced on GitHub [1]. © IFIP International Federation for Information Processing 2025.

关键词： Data accuracy

来源：评论

学校读者我要写书评

暂无评论

A Modular-based Strategy for Mitigating Gradient Conflicts in Simultaneous Speech Translation

A Modular-based Strategy for Mitigating Gradient Conflicts i...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Liu, Xiaoqian Du, Yangfan Wang, Jianjin Ge, Yuan Xu, Chen Xiao, Tong Chen, Guocheng Zhu, Jingbo School of Computer Science and Engineering Northeastern University Shenyang China College of Computer Science and Technology Harbin Engineering University Harbin China NiuTrans Research Shenyang China

ISBN: (纸本)9798350368741

Simultaneous Speech Translation (SimulST) involves generating target language text while continuously processing streaming speech input, presenting significant real-time challenges. Multi-task learning is often employed to enhance SimulST performance but introduces optimization conflicts between primary and auxiliary tasks, potentially compromising overall efficiency. The existing model-level conflict resolution methods are not well-suited for this task which exacerbates inefficiencies and leads to high GPU memory consumption. To address these challenges, we propose a Modular Gradient Conflict Mitigation (MGCM) strategy that detects conflicts at a finer-grained modular level and resolves them utilizing gradient projection. Experimental results demonstrate that MGCM significantly improves SimulST performance, particularly under medium and high latency conditions, achieving a 0.68 BLEU score gain in offline tasks. Additionally, MGCM reduces GPU memory consumption by over 95% compared to other conflict mitigation methods, establishing it as a robust solution for SimulST tasks. © 2025 IEEE.

关键词： GPU Memory Efficiency Gradient Conflict Mitigation Multi-task Learning Simultaneous Speech Translation

来源：评论

学校读者我要写书评

暂无评论

An Overview of Wireless Charging Systems for Electric Vehicles: From Historical Evolution to Modern Applications 26

An Overview of Wireless Charging Systems for Electric Vehicl...

引用

26th International Conference on Industrial technology, ICIT 2025

作者： Liu, Ruoning Xu, Zhanguo Chen, Jiangchuan Wang, Yun School of Transportation Science and Engineering Harbin Institute of Technology Harbin China

ISBN: (纸本)9798331521950

Wireless charging systems have emerged as a convenient and efficient solution for powering electronic devices and electric vehicles without the requirement for physical connectors, gaining significant attention in recent years. This paper reviews the current development of wireless charging technology, including its background, main classifications, and common charging methods, such as inductive power transfer and coupled magnetic resonance. Newer approaches, key challenges impacting power transmission efficiency, and corresponding mitigation strategies are also examined. Safety considerations and environmental impacts are evaluated to understand their influence on technology adoption. Finally, future trends and research priorities are outlined, proposing innovations needed to address current limitations and enhance wireless power transmission capabilities. This review aims to provide a comprehensive overview of wireless charging systems, aiding researchers and practitioners in understanding recent advancements and contributing to the development of next-generation solutions. © 2025 IEEE.

关键词： Inductive power transmission

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：