检索结果-内蒙古大学图书馆

International Conference on Acoustics, Speech, and Signal Processing (ICASSP)

作者： Hao Ma Zhiyuan Peng Xu Li Yukai Li Mingjie Shao Qiuqiang Kong Ju Liu School of Information Science and Engineering Shandong University Qingdao China Department of Computer Science North Carolina State University North Carolina USA ARC Lab Tencent PCG Key Laboratory of System and Control AMSS Chinese Academy of Sciences Beijing China The Chinese University of Hong Kong China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Language-queried target sound extraction (TSE) aims to extract specific sounds from mixtures based on language queries. Traditional fully-supervised training schemes require extensively annotated parallel audio-text data, which are labor-intensive. We introduce a parallel-data-free training scheme, requiring only unlabelled audio clips for TSE model training by utilizing the contrastive language-audio pre-trained model (CLAP). In a vanilla parallel-data-free training stage, target audio is encoded using the pre-trained CLAP audio encoder to form a condition embedding, while during testing, user language queries are encoded by CLAP text encoder as the condition embedding. This vanilla approach assumes perfect alignment between text and audio embeddings, which is unrealistic. Two major challenges arise from training-testing mismatch: the persistent modality gap between text and audio and the risk of overfitting due to the exposure of rich acoustic details in target audio embedding during training. To address this, we propose a retrieval-augmented strategy. Specifically, we create an embedding cache using audio captions generated by a large language model (LLM). During training, target audio embeddings retrieve text embeddings from this cache to use as condition embeddings, ensuring consistent modalities between training and testing and eliminating information leakage. Extensive experiment results show that our retrieval-augmented approach achieves consistent and notable performance improvements over existing state-of-the-art with better generalizability.

关键词： Training Large language models Training data Signal processing Information leakage Acoustics Data models Data mining Speech processing Overfitting

来源：评论

学校读者我要写书评

暂无评论

On Completeness of SDP-Based Barrier Certificate Synthesis over Unbounded Domains

arXiv

引用

arXiv 2023年

作者： Wu, Hao Feng, Shenghua Gan, Ting Wang, Jie Xia, Bican Zhan, Naijun State Key Lab. of Computer Science Institute of Software University of Chinese Academy of Sciences Beijing China Zhongguancun Laboratory Beijing China School of Computer Science Wuhan University China Academy of Mathematics and Systems Science CAS Beijing China School of Mathematical Sciences Peking University Beijing China School of Computer Science Peking University Beijing China

Barrier certificates, serving as differential invariants that witness system safety, play a crucial role in the verification of cyber-physical systems (CPS). Prevailing computational methods for synthesizing barrier certificates are based on semidefinite programming (SDP) by exploiting Putinar Positivstellensatz. Consequently, these approaches are limited by the Archimedean condition, which requires all variables to be bounded, i.e., systems are defined over bounded domains. For systems over unbounded domains, unfortunately, existing methods become incomplete and may fail to identify potential barrier certificates. In this paper, we address this limitation for the unbounded cases. We first give a complete characterization of polynomial barrier certificates by using homogenization, a recent technique in the optimization community to reduce an unbounded optimization problem to a bounded one. Furthermore, motivated by this formulation, we introduce the definition of homogenized systems and propose a complete characterization of a family of non-polynomial barrier certificates with more expressive power. Experimental results demonstrate that our two approaches are more effective while maintaining a comparable level of efficiency. Copyright © 2023, The Authors. All rights reserved.

关键词： Homogenization method

来源：评论

学校读者我要写书评

暂无评论

PSTNet: Enhanced Polyp Segmentation with Multi-scale Alignment and Frequency Domain Integration

arXiv

引用

arXiv 2024年

作者： Xu, Wenhao Xu, Rongtao Wang, Changwei Li, Xiuli Xu, Shibiao Guo, Li School of Artificial Intelligence Beijing University of Posts and Telecommunications Beijing100876 China The State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation Chinese Academy of Sciences Beijing China Jinan250013 China Shandong Provincial Key Laboratory of Computer Networks Shandong Fundamental Research Center for Computer Science Jinan China AI Lab Deepwise Healthcare Beijing100080 China

Accurate segmentation of colorectal polyps in colonoscopy images is crucial for effective diagnosis and management of colorectal cancer (CRC). However, current deep learning-based methods primarily rely on fusing RGB information across multiple scales, leading to limitations in accurately identifying polyps due to restricted RGB domain information and challenges in feature misalignment during multi-scale aggregation. To address these limitations, we propose the Polyp Segmentation Network with Shunted Transformer (PSTNet), a novel approach that integrates both RGB and frequency domain cues present in the images. PSTNet comprises three key modules: the Frequency Characterization Attention Module (FCAM) for extracting frequency cues and capturing polyp characteristics, the Feature Supplementary Alignment Module (FSAM) for aligning semantic information and reducing misalignment noise, and the Cross Perception localization Module (CPM) for synergizing frequency cues with high-level semantics to achieve efficient polyp segmentation. Extensive experiments on challenging datasets demonstrate PSTNet’s significant improvement in polyp segmentation accuracy across various metrics, consistently outperforming state-of-the-art methods. The integration of frequency domain cues and the novel architectural design of PSTNet contribute to advancing computer-assisted polyp segmentation, facilitating more accurate diagnosis and management of CRC. © 2024, CC BY.

关键词： Frequency domain analysis

来源：评论

学校读者我要写书评

暂无评论

Focal volume optics for composite structuring in transparent solids

引用

International Journal of Extreme Manufacturing 2025年第1期7卷 357-366页

作者： Bo Zhang Zhuo Wang Dezhi Tan Min Gu Yuanzheng Yue Jianrong Qiu State Key Laboratory of Modern Optical Instrumentation College of Optical Science and EngineeringZhejiang UniversityHangzhou 310027People’s Republic of China Zhejiang Lab Hangzhou 311100People’s Republic of China School of Materials Science and Engineering Zhejiang UniversityHangzhou 310027People’s Republic of China Institute of Photonic Chips University of Shanghai for Science and TechnologyShanghaiPeople’s Republic of China Centre for Artificial-Intelligence Nanophotonics School of Optical-Electrical and Computer EngineeringUniversity of Shanghai for Science and TechnologyShanghaiPeople’s Republic of China Department of Chemistry and Bioscience Aalborg University9220 AalborgDenmark

Achieving high-level integration of composite micro-nano structures with different structural characteristics through a minimalist and universal process has long been the goal pursued by advanced manufacturing research but is rarely explored due to the absence of instructive ***,we revealed a controllable ultrafast laser-induced focal volume light field and experimentally succeeded in highly efficient one-step composite structuring in multiple transparent solids.A pair of spatially coupled twin periodic structures reflecting light distribution in the focal volume are simultaneously created and independently tuned by engineering ultrafast laser-matter *** demonstrated that the generated composite micro-nano structures are applicable to multi-dimensional information integration,nonlinear diffractive elements,and multi-functional optical *** work presents the experimental verification of highly universal all-optical fabrication of composite micro-nano structures with independent controllability in multiple degrees of freedom,expands the current cognition of ultrafast laser-based material modification in transparent solids,and establishes a new scientific aspect of strong-field optics,namely,focal volume optics for composite structuring transparent solids.

关键词： ultrafast laser focal volume light field composite structuring transparent solids advanced manufacturing

来源：评论

学校读者我要写书评

暂无评论

BroadCAM: Outcome-agnostic Class Activation Mapping for Small-scale Weakly Supervised Applications

arXiv

引用

arXiv 2023年

作者： Lin, Jiatai Han, Guoqiang Xu, Xuemiao Liang, Changhong Wong, Tien-Tsin Chen, C.L. Philip Liu, Zaiyi Han, Chu The School of Computer Science and Engineering South China University of Technology Guangdong Guangzhou510006 China Guangdong Provincial Key Laboratory of Artificial Intelligence in Medical Image Analysis and Application Guangzhou510080 China Southern Medical University Guangzhou510080 China School of Computer Science and Engineering South China University of Technology State Key Laboratory of Subtropical Building Science Ministry of Education Key Laboratory of Big Data and Intelligent Robot Guangdong Provincial Key Lab of Computational Intelligence and Cyberspace Information China Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong

Class activation mapping (CAM), a visualization technique for interpreting deep learning models, is now commonly used for weakly supervised semantic segmentation (WSSS) and object localization (WSOL). It is the weighted aggregation of the feature maps by activating the high class-relevance ones. Current CAM methods achieve it relying on the training outcomes, such as predicted scores (forward information), gradients (backward information), etc. However, when with small-scale data, unstable training may lead to less effective model outcomes and generate unreliable weights, finally resulting in incorrect activation and noisy CAM seeds. In this paper, we propose an outcome-agnostic CAM approach, called BroadCAM, for small-scale weakly supervised applications. Since broad learning system (BLS) is independent to the model learning, BroadCAM can avoid the weights being affected by the unreliable model outcomes when with small-scale data. By evaluating BroadCAM on VOC2012 (natural images) and BCSS-WSSS (medical images) for WSSS and OpenImages30k for WSOL, BroadCAM demonstrates superior performance than existing CAM methods with small-scale data (less than 5%) in different CNN architectures. It also achieves SOTA performance with large-scale training data. Extensive qualitative comparisons are conducted to demonstrate how BroadCAM activates the high class-relevance feature maps and generates reliable CAMs when with small-scale training data. © 2023, CC BY.

关键词： Learning systems

来源：评论

学校读者我要写书评

暂无评论

QoE-Aware Latency Optimization in Semantic Transmission Empowered Edge Assisted AR

引用

IEEE Transactions on Vehicular Technology 2025年

作者： Li, Yang Wu, Yuan Xing, Huanlai Feng, Li Wang, Tian Jia, Weijia Southwest Jiaotong University School of Computer and Artificial Intelligent Chengdu611756 China University of Macau State Key Laboratory of Internet of Things for Smart City Department of Computer and Information Science Macau China Beijing Normal University Institute of Artificial Intelligence and Future Networks Zhuhai519087 China Guangdong Key Lab of AI and Multi-Modal Data Processing BNU-HKBU United International College Zhuhai519087 China

Augmented reality (AR) enhances human perception by overlaying digital information in the real environment and can be applied to various fields such as education, healthcare, and industry. To relieve the computing pressure of resourcelimited AR devices, edge assisted AR has attracted significant attention in recent years. It enables AR devices to offload the computation-intensive environment analysis tasks to the edge server and to download the point-based 3D virtual objects for rendering. Although this method can effectively reduce the processing latency, it may introduce a long transmission latency. To address this issue, this work proposes a semantic communication empowered edge assisted AR to reduce the quality of experience (QoE)-aware latency. To this end, the AR devices first utilize an image encoder to extract semantic features of source images. Due to the importance difference of extracted features and limited communication resources, the feature selection and bandwidth allocation are explored to improve transmission efficiency. In addition, when the edge server completes the generation of pointbased 3D virtual objects, it employs a transformer-based encoder to mask a proportion of points and learn a high-level representation of unmasked points. In this way, low-complexity semantic information of 3D objects is sent to the AR devices, which can reduce the transmission overhead of 3D objects. Moreover, the mask ratio of the point encoder is further optimized to balance the reconstructed quality and transmission latency of 3D objects. Therefore, this work aims to minimize the overall QoE-aware latency of edge assisted AR by jointly optimizing the semantic feature selection, mask ratio, and bandwidth allocation. A convexbased improved Harris Hawks optimization algorithm is designed to solve the formulated problem. Experimental results demonstrate that our semantic communication empowered edge assisted AR can effectively reduce the overall latency and outperform

关键词： Augmented reality

来源：评论

学校读者我要写书评

暂无评论

Pano-NeRF: Synthesizing High Dynamic Range Novel Views with Geometry from Sparse Low Dynamic Range Panoramic Images

arXiv

引用

arXiv 2023年

作者： Lu, Zhan Zheng, Qian Shi, Boxin Jiang, Xudong School of Electrical and Electronic Engineering Nanyang Technological University Singapore College of Computer Science and Technology Zhejiang University China The State Key Lab of Brain-Machine Intelligence Zhejiang University China National Key Laboratory for Multimedia Information Processing School of Computer Science Peking University China National Engineering Research Center of Visual Technology School of Computer Science Peking University China

Panoramic imaging research on geometry recovery and High Dynamic Range (HDR) reconstruction becomes a trend with the development of Extended Reality (XR). Neural Radiance Fields (NeRF) provide a promising scene representation for both tasks without requiring extensive prior data. However, in the case of inputting sparse Low Dynamic Range (LDR) panoramic images, NeRF often degrades with under-constrained geometry and is unable to reconstruct HDR radiance from LDR inputs. We observe that the radiance from each pixel in panoramic images can be modeled as both a signal to convey scene lighting information and a light source to illuminate other pixels. Hence, we propose the irradiance fields from sparse LDR panoramic images, which increases the observation counts for faithful geometry recovery and leverages the irradiance-radiance attenuation for HDR reconstruction. Extensive experiments demonstrate that the irradiance fields outperform state-of-the-art methods on both geometry recovery and HDR reconstruction and validate their effectiveness. Furthermore, we show a promising byproduct of spatially-varying lighting estimation. The code is available at https://***/Lu-Zhan/Pano-NeRF. Copyright © 2023, The Authors. All rights reserved.

关键词： Recovery

来源：评论

学校读者我要写书评

暂无评论

A 52-Gb/s Sub-1-pJ/bit PAM4 Receiver in 40-nm CMOS for Low-Power Interconnects

IEEE Open Journal of Circuits and Systems

引用

IEEE Open Journal of Circuits and Systems 2021年 2卷 46-55页

作者： Wang, Can Wang, Li Zhang, Zhao Mahmoudabadi, Milad Kalantari Shi, Weimin Yue, C. Patrick Department of Electronic and Computer Engineering HKUST-Qualcomm Optical Wireless Lab Hong Kong University of Science and Technology Hong Kong State Key Laboratory of Superlattices and Microstructures Institute of Semiconductors Chinese Academy of Sciences Beijing China HKUST Shenzhen Research Institute Shenzhen518057 China

This article presents a quarter-rate source-synchronous PAM-4 receiver for energy-efficient chip-to-module communication. A novel single-stage multiple peaking continuous-time linear equalizer (MP-CTLE) using feedback enabled multiple peaking scheme for both high-frequency equalization (HF-EQ) and low-frequency equalization (LF-EQ) is proposed to improve the BER performance and overall energy efficiency. The LF-EQ of the MP-CTLE eliminates the need for many DFE taps in SR/VSR applications to save power and area. A 1-tap feedforward equalizer (FFE) is used to further compensate for the high-frequency loss. We also use a ring oscillator based wide bandwidth phase-locked loop (WBW-PLL) as the multiphase clock generator (MPCG) in the clock and data recovery loop to save power with acceptable phase accuracy. Fabricated in 40-nm CMOS technology, the prototype receiver chip achieves error-free operation up to 52 Gb/s PAM-4 with superior bit efficiency of 0.126pJ/bit/s/dB while compensating 7.3-dB channel loss at 13GHz. With the proposed single-stage MP-CTLE, the receiver extends the error-free operation from PRBS-7 to PRBS-9. The BER bathtub curve at 10-6 is improved from 0.018 UI to around 0.1 UI. © 2020 IEEE.

关键词： Pulse amplitude modulation

来源：评论

学校读者我要写书评

暂无评论

CoFInAl: Enhancing Action Quality Assessment with Coarse-to-Fine Instruction Alignment

arXiv

引用

arXiv 2024年

作者： Zhou, Kanglei Li, Junlin Cai, Ruizhi Wang, Liyuan Zhang, Xingxing Liang, Xiaohui Department of Computer Science and Technology Institute for AI BNRist Center Tsinghua-Bosch Joint ML Center THBI Lab Tsinghua University China Zhongguancun Laboratory China State Key Laboratory of Virtual Reality Technology and Systems Beihang University China China Three Gorges University China

Action Quality Assessment (AQA) is pivotal for quantifying actions across domains like sports and medical care. Existing methods often rely on pre-trained backbones from large-scale action recognition datasets to boost performance on smaller AQA datasets. However, this common strategy yields suboptimal results due to the inherent struggle of these backbones to capture the subtle cues essential for AQA. Moreover, fine-tuning on smaller datasets risks overfitting. To address these issues, we propose Coarse-to-Fine Instruction Alignment (CoFInAl). Inspired by recent advances in large language model tuning, CoFInAl aligns AQA with broader pre-trained tasks by reformulating it as a coarse-to-fine classification task. Initially, it learns grade prototypes for coarse assessment and then utilizes fixed sub-grade prototypes for fine-grained assessment. This hierarchical approach mirrors the judging process, enhancing interpretability within the AQA framework. Experimental results on two long-term AQA datasets demonstrate CoFInAl achieves state-of-the-art performance with significant correlation gains of 5.49% and 3.55% on Rhythmic Gymnastics and Fis-V, respectively. Copyright © 2024, The Authors. All rights reserved.

关键词： Large datasets

来源：评论

学校读者我要写书评

暂无评论

Tackling Noisy labels with Network Parameter Additive Decomposition

arXiv

引用

arXiv 2024年

作者： Wang, Jingyi Xia, Xiaobo Lan, Long Wu, Xinghao Yu, Jun Yang, Wenjing Han, Bo Liu, Tongliang The Department of Intelligent Data Science College of Computer Science and Technology National University of Defense Technology Changsha410073 China The Trustworthy Machine Learning Lab School of Computer Science Faculty of Engineering University of Sydney DarlingtonNSW2008 Australia The State Key Laboratory of Virtual Reality Technology and Systems School of Computer Science and Engineering Beihang University Beijing100191 China The Department of Automation University of Science and Technology of China Hefei230026 China The Department of Computer Science Hong Kong Baptist University Hong Kong

Given data with noisy labels, over-parameterized deep networks suffer overfitting mislabeled data, resulting in poor generalization. The memorization effect of deep networks shows that although the networks have the ability to memorize all noisy data, they would first memorize clean training data, and then gradually memorize mislabeled training data. A simple and effective method that exploits the memorization effect to combat noisy labels is early stopping. However, early stopping cannot distinguish the memorization of clean data and mislabeled data, resulting in the network still inevitably overfitting mislabeled data in the early training stage. In this paper, to decouple the memorization of clean data and mislabeled data, and further reduce the side effect of mislabeled data, we perform additive decomposition on network parameters. Namely, all parameters are additively decomposed into two groups, i.e. , parameters w are decomposed as w = σ + γ. Afterward, the parameters σ are considered to memorize clean data, while the parameters γ are considered to memorize mislabeled data. Benefiting from the memorization effect, the updates of the parameters σ are encouraged to fully memorize clean data in early training, and then discouraged with the increase of training epochs to reduce interference of mislabeled data. The updates of the parameters γ are the opposite. In testing, only the parameters σ are employed to enhance generalization. Extensive experiments on both simulated and real-world benchmarks confirm the superior performance of our method. The implementation is available at: https://***/wangjingyi9924/TNLPAD. Copyright © 2024, The Authors. All rights reserved.

关键词： Benchmarking

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：