检索结果-内蒙古大学图书馆

Learning to Distill Global Representation for Sparse-View CT

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Zilong Ma, Chenglong Chen, Jie Zhang, Junping Shan, Hongming Shanghai Key Lab of Intelligent Information Processing School of Computer Science Fudan University Shanghai200433 China Institute of Science and Technology for Brain-inspired Intelligence and MOE Frontiers Center for Brain Science Fudan University Shanghai200433 China Shanghai Center for Brain Science and Brain-inspired Technology Shanghai200031 China

Sparse-view computed tomography (CT)—using a small number of projections for tomographic reconstruction—enables much lower radiation dose to patients and accelerated data acquisition. The reconstructed images, however, suffer from strong artifacts, greatly limiting their diagnostic value. Current trends for sparse-view CT turn to the raw data for better information recovery. The resultant dual-domain methods, nonetheless, suffer from secondary artifacts, especially in ultra-sparse view scenarios, and their generalization to other scanners/protocols is greatly limited. A crucial question arises: have the image post-processing methods reached the limit? Our answer is not yet. In this paper, we stick to image post-processing methods due to great flexibility and propose global representation(GloRe) distillation framework for sparse-view CT, termed GloReDi. First, we propose to learn GloRe with Fourier convolution, so each element in GloRe has an image-wide receptive field. Second, unlike methods that only use the full-view images for supervision, we propose to distill GloRe from intermediate-view reconstructed images that are readily availab.e but not explored in previous literature. The success of GloRe distillation is attributed to two key components: representation directional distillation to align the GloRe directions, and band-pass-specific contrastive distillation to gain clinically important details. Extensive experiments demonstrate the superiority of the proposed GloReDi over the state-of-the-art methods, including dual-domain ones. The source code is availab.e at https://***/longzilicart/GloReDi. Copyright © 2023, The Authors. All rights reserved.

关键词： Distillation

Prompt-In-Prompt Learning for Universal Image Restoration

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Li, Zilong Lei, Yiming Ma, Chenglong Zhang, Junping Shan, Hongming Shanghai Key Lab of Intelligent Information Processing School of Computer Science Fudan University Shanghai200433 China Institute of Science and Technology for Brain-inspired Intelligence and MOE Frontiers Center for Brain Science Fudan University Shanghai200433 China Shanghai Center for Brain Science and Brain-inspired Technology Shanghai200031 China

Image restoration, which aims to retrieve and enhance degraded images, is fundamental across a wide range of applications. While conventional deep learning approaches have notably improved the image quality across various tasks, they still suffer from (i) the high storage cost needed for various task-specific models and (ii) the lack of interactivity and flexibility, hindering their wider application. Drawing inspiration from the pronounced success of prompts in both linguistic and visual domains, we propose novel Prompt-In-Prompt learning for universal image restoration, named PIP. First, we present two novel prompts, a degradation-aware prompt to encode high-level degradation knowledge and a basic restoration prompt to provide essential low-level information. Second, we devise a novel prompt-to-prompt interaction module to fuse these two prompts into a universal restoration prompt. Third, we introduce a selective prompt-to-feature interaction module to modulate the degradation-related feature. By doing so, the resultant PIP works as a plug-and-play module to enhance existing restoration models for universal image restoration. Extensive experimental results demonstrate the superior performance of PIP on multiple restoration tasks, including image denoising, deraining, dehazing, deblurring, and low-light enhancement. Remarkably, PIP is interpretable, flexible, efficient, and easy-to-use, showing promising potential for real-world applications. The code is availab.e at https://***/longzilicart/pip_universal. Copyright © 2023, The Authors. All rights reserved.

关键词： Image reconstruction

Dual Interactive Graph Convolutional Networks for Hyperspectral Image Classification

学校读者我要写书评

暂无评论

IEEE Transactions on Geoscience and Remote Sensing 2022年 60卷 1页

作者： Wan, Sheng Pan, Shirui Zhong, Ping Chang, Xiaojun Yang, Jian Gong, Chen Pca Laboratory Key Lab. of Intelligent Percept. and Syst. for High-Dimensional Information of Ministry of Education Jiangsu Key Laboratory of Image and Video Understanding for Social Security School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing210094 China Faculty of Information Technology Monash University ClaytonVIC3800 Australia National Key Laboratory of Science and Technology on Atr National University of Defense Technology Changsha410073 China Pca Laboratory Key Lab. of Intelligent Percept. and Syst. for High-Dimensional Information of Ministry of Education Nanjing University of Science and Technology Nanjing210094 China Department of Computing Hong Kong Polytechnic University Hong Kong Hong Kong

Recently, graph convolutional network (GCN) has progressed significantly and gained increasing attention in hyperspectral image (HSI) classification due to its impressive representation power. However, existing GCN-based methods do not give full consideration to the multiscale spatial information, since the convolution operations are governed by fixed neighborhood. As a result, their performances can be limited, particularly in the regions with diverse land cover appearances. In this article, we develop a new dual interactive GCN (DIGCN) which introduces the dual GCN branches to capture spatial information at different scales. More significantly, the dual interactive module is embedded across the GCN branches, so that the correlation of multiscale spatial information can be leveraged to refine the graph information. To be concrete, the edge information contained in one GCN branch can be refined by incorporating the feature representations from the other branch. Analogously, improved feature representations can be generated in one GCN branch by fusing the edge information from the other branch. As such, the refined graph information can help enhance the representation power of the model. Furthermore, to avoid the negative effects of the manually constructed graph, our proposed model adaptively learns a discriminative region-induced graph, which also accelerates the convolution operation. We comprehensively evaluate the proposed method on four commonly used HSI benchmark data sets, and the state-of-the-art results can be achieved when compared with several typical HSI classification methods. © 1980-2012 IEEE.

关键词： Convolution

Array optimization for MIMO radar based on harmony search mechanism 11

学校读者我要写书评

暂无评论

Array optimization for MIMO radar based on harmony search me...

11th International Conference on information technology in Medicine and Education, ITME 2021

作者： Jiang, Ling Jiang, Yi Shi, Guimin Xiao, Zhongjie Wuyi University School of Mathematics and Computer Wuyishan China The Key Lab. of Cogn. Comp. and Intelligent Information Processing of Fujian Education Institutions Wuyishan China Digital Fujian Tourism Big Data Institute Wuyishan China Big Data Development Administration of Anji County Huzhou China

ISBN: (纸本)9781665406796

In order to overcome the premarure risk of differential evolution algorithm, harmony search mechanism is introduced to optimize the pattern synthesis of multiple input and multiple output radar. Firstly, the memory of harmony search receives the optimizing results of differential evolution algorithm. And then it disturbs the local best to achieve better global optimization results. The advantage of novel method than standard differential evolution is tested with benchmark function while it can also maintain the diversity of population. What's more, several experiments are conducted to show that the optimal peak side lobe level and convergence performance have been achieved better through the proposed algorithm for multiple input and multiple output radar. © 2021 IEEE.

关键词： MIMO radar

Optimizing Neighborhood Area Network Topology for Enhanced Communication in Smart Grid

学校读者我要写书评

暂无评论

IEEE Transactions on Consumer Electronics 2025年

作者： Li, Jian-Hui Ding, Yuemin Yu, Zu-Guo Zhang, Weizhe Ren, Honglin Tian, Yu-Chu Xiangtan University National Center for Applied Mathematics in Hunan Key Laboratory of Intelligent Computing and Information Processing of the Ministry of Education of China Xiangtan University Hunan 411105 China Queensland University of Technology School of Computer Science Brisbane QLD Australia University of Navarra Ikerbasque Foundation Tecnun School of Engineering San Sebastián Spain Bilbao Spain Harbin Institute of Technology School of Computer Science and Technology Shenzhen China

In modern power grids, smart meters, as a type of consumer electronics, act as metering and communication nodes, which form Neighborhood Area Networks (NANs) spanning electricity distribution networks from substations to energy producers and consumers. In smart meter communication, wireless technologies have been widely adopted due to their inherent advantages of easier installation, flexibility, and low cost. For achieving low-latency communications in NANs, it is crucial to optimize the existing topology to enhance communication efficiency. However, optimizing the NAN topology is a challenging task due to the computational infeasibility of obtaining the optimal solution. To tackle this challenge, topology optimization in NANs is first formulated into a combinatorial optimization problem. After that, a heuristic solution driven by complex network theory is presented to solve the formulated optimization problem. The proposed heuristic solution consists of an effective network rewiring scheme and a long-range edge connection strategy. The rewiring optimizes the topology of wireless NANs to reduce local network length, whilst the long-range edge connection best pairs a small number of NAN nodes to reduce global network length. Numerical studies are conducted on diverse case studies to demonstrate the presented solution to enhance communication efficiency in smart grid NANs, outperforming the benchmark schemes. © 1975-2011 IEEE.

关键词： communication efficiency neighborhood area network network rewiring network topological optimization Network topology smart grid smart meter

Learning Music-Dance Representations through Explicit-Implicit Rhythm Synchronization

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Yu, Jiashuo Pu, Junfu Cheng, Ying Feng, Rui Shan, Ying The School of Computer Science Shanghai Key Lab of Intelligent Information Processing Shanghai Collaborative Innovation Center of Intelligent Visual Computing Fudan University Shanghai200438 China The Academy for Engineering and Technology Fudan University Shanghai200438 China The Applied Research Center PCG Tencent Shenzhen518000 China

Although audio-visual representation has been proven to be applicable in many downstream tasks, the representation of dancing videos, which is more specific and always accompanied by music with complex auditory contents, remains challenging and uninvestigated. Considering the intrinsic alignment between the cadent movement of the dancer and music rhythm, we introduce MuDaR, a novel Music-Dance Representation learning framework to perform the synchronization of music and dance rhythms both in explicit and implicit ways. Specifically, we derive the dance rhythms based on visual appearance and motion cues inspired by the music rhythm analysis. Then the visual rhythms are temporally aligned with the music counterparts, which are extracted by the amplitude of sound intensity. Meanwhile, we exploit the implicit coherence of rhythms implied in audio and visual streams by contrastive learning. The model learns the joint embedding by predicting the temporal consistency between audio-visual pairs. The music-dance representation, together with the capability of detecting audio and visual rhythms, can further be applied to three downstream tasks: (a) dance classification, (b) music-dance retrieval, and (c) music-dance retargeting. Extensive experiments demonstrate that our proposed framework outperforms other self-supervised methods by a large margin. © 2022, CC BY.

关键词： Supervised learning

Exploiting Completeness and Uncertainty of Pseudo lab.ls for Weakly Supervised Video Anomaly Detection

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhang, Chen Li, Guorong Qi, Yuankai Wang, Shuhui Qing, Laiyun Huang, Qingming Yang, Ming-Hsuan State Key Laboratory of Information Security Institute of Information Engineering CAS China School of Cyber Security University of Chinese Academy of Sciences China School of Computer Science and Technology University of Chinese Academy of Sciences China Australian Institute for Machine Learning The University of Adelaide Australia Key Laboratory of Intelligent Information Processing Institute of Computing Technology CAS China Peng Cheng Laboratory China University of California Merced United States

Weakly supervised video anomaly detection aims to identify abnormal events in videos using only video-level lab.ls. Recently, two-stage self-training methods have achieved significant improvements by self-generating pseudo lab.ls and self-refining anomaly scores with these lab.ls. As the pseudo lab.ls play a crucial role, we propose an enhancement framework by exploiting completeness and uncertainty properties for effective self-training. Specifically, we first design a multi-head classification module (each head serves as a classifier) with a diversity loss to maximize the distribution differences of predicted pseudo lab.ls across heads. This encourages the generated pseudo lab.ls to cover as many abnormal events as possible. We then devise an iterative uncertainty pseudo lab.l refinement strategy, which improves not only the initial pseudo lab.ls but also the updated ones obtained by the desired classifier in the second stage. Extensive experimental results demonstrate the proposed method performs favorably against state-of-the-art approaches on the UCF-Crime, TAD, and XD-Violence benchmark datasets. Copyright © 2022, The Authors. All rights reserved.

关键词： Anomaly detection

DualGOFiller: A Dual-Channel Graph Neural Network with Contrastive Learning for Enhancing Function Prediction in Partially Annotated Proteins 29th

学校读者我要写书评

暂无评论

DualGOFiller: A Dual-Channel Graph Neural Network with Cont...

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

作者： Wang, Shaojun Liu, Hancheng Zhai, Weiqi Zhu, Shanfeng Institute of Science and Technology for Brain-Inspired Intelligence Fudan University Shanghai200433 China MOE Frontiers Center for Brain Science Fudan University Shanghai200433 China Ministry of Education Shanghai200433 China Shanghai Key Lab of Intelligent Information Processing and Shanghai Institute of Artificial Intelligence Algorithm Fudan University Shanghai200433 China Zhangjiang Fudan International Innovation Center Shanghai200433 China

ISBN: (纸本)9783031902512

Protein function prediction remains a critical and challenging task in the field of bioinformatics. Despite various existing computational methods, most focus on predicting functions for novel proteins, often overlooking those with incomplete annotations. Predicting missing Gene Ontology (GO) terms for these partially annotated proteins is also important for gaining a comprehensive understanding of their biological roles. To address this gap, we introduce DualGOFiller, a dual-channel graph neural network (GNN) model, which is a novel method specifically focused on predicting missing GO terms for partially annotated proteins. DualGOFiller integrates heterogeneous (protein-GO bipartite graph) and homogeneous (protein-protein interaction network and GO directed acyclic graph) network information, enriched with attribute data from protein sequences and GO term descriptions. It further employs graph contrastive learning (GCL) to strengthen the alignment across different views. Experiments demonstrate the superiority of DualGOFiller over state-of-the-art methods, including traditional recommendation system approaches and several protein function prediction methods. The ablation study further confirms the significant contributions of key component in DualGOFiller for boosting predictive performance. The source code is availab.e at https://***/Zhulab.Fudan/DualGOFiller. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Contrastive Learning

Pushing Auto-regressive Models for 3D Shape Generation at Capacity and Scalab.lity

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Qian, Xuelin Wang, Yu Luo, Simian Zhang, Yinda Tai, Ying Zhang, Zhenyu Wang, Chengjie Xue, Xiangyang Zhao, Bo Huang, Tiejun Wu, Yunsheng Fu, Yanwei School of Data Science Fudan ISTBI—ZJNU Algorithm Centre for Brain-inspired Intelligence Fudan University China Institute for Interdisciplinary Information Sciences Tsinghua University China Google United States School of Intelligence Science and Technology Nanjing University Suzhou Campus China Tencent Youtu lab Shenzhen China Beijing Academy of Artificial Intelligence China School of Computer Science Shanghai Key Lab of Intelligent Information Processing Fudan University China

Auto-regressive models have achieved impressive results in 2D image generation by modeling joint distributions in grid space. In this paper, we extend auto-regressive models to 3D domains, and seek a stronger ability of 3D shape generation by improving auto-regressive models at capacity and scalab.lity simultaneously. Firstly, we leverage an ensemble of publicly availab.e 3D datasets to facilitate the training of large-scale models. It consists of a comprehensive collection of approximately 900, 000 objects, with multiple properties of meshes, points, voxels, rendered images, and text captions. This diverse lab.led dataset, termed Objaverse-Mix, empowers our model to learn from a wide range of object variations. For data processing, we employ four machines with 64-core CPUs and 8 A100 GPUs each over four weeks, utilizing nearly 100TB of storage due to process complexity. And the dataset is on https://***/datasets/BAAI/Objaverse-MIX. However, directly applying 3D auto-regression encounters critical challenges of high computational demands on volumetric grids and ambiguous auto-regressive order along grid dimensions, resulting in inferior quality of 3D shapes. To this end, we then present a novel framework Argus3D in terms of capacity. Concretely, our approach introduces discrete representation learning based on a latent vector instead of volumetric grids, which not only reduces computational costs but also preserves essential geometric details by learning the joint distributions in a more tractable order. The capacity of conditional generation can thus be realized by simply concatenating various conditioning inputs to the latent vector, such as point clouds, categories, images, and texts. In addition, thanks to the simplicity of our model architecture, we naturally scale up our approach to a larger model with an impressive 3.6 billion parameters, further enhancing the quality of versatile 3D generation. Extensive experiments on four generation tasks demons

关键词： Scalab.lity