检索结果-内蒙古大学图书馆

Contrastive Continual learning with Importance Sampling and Prototype-Instance Relation Distillation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Li, Jiyong Azizov, Dilshod Li, Yang Liang, Shangsong School of Computer Science and Engineering Sun Yat-sen University China Guangdong Key Laboratory of Big Data Analysis and Processing Guangzhou China Department of Machine Learning Mohamed bin Zayed University of Artificial Intelligence United Arab Emirates AI Thrust Information Hub The Hong Kong University of Science and Technology Guangzhou China Department of CSE The Hong Kong University of Science and Technology China

Recently, because of the high-quality representations of contrastive learning methods, rehearsal-based contrastive continual learning has been proposed to explore how to continually learn transferable representation embeddings to avoid the catastrophic forgetting issue in traditional continual settings. Based on this framework, we propose Contrastive Continual learning via Importance Sampling (CCLIS) to preserve knowledge by recovering previous data distributions with a new strategy for Replay Buffer Selection (RBS), which minimize estimated variance to save hard negative samples for representation learning with high quality. Furthermore, we present the Prototype-instance Relation Distillation (PRD) loss, a technique designed to maintain the relationship between prototypes and sample representations using a self-distillation process. Experiments on standard continual learning benchmarks reveal that our method notably outperforms existing baselines in terms of knowledge preservation and thereby effectively counteracts catastrophic forgetting in online contexts. The code is available at https://***/lijy373/CCLIS. Copyright © 2024, The Authors. All rights reserved.

关键词： Distillation

Weakly supervised image segmentation for defect-based grading of fresh produce

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Knott, Manuel Odion, Divinefavour Sontakke, Sameer Karwa, Anup Defraeye, Thijs Empa Swiss Federal Laboratories for Materials Science and Technology Laboratory for Biomimetic Membranes and Textiles St. Gallen Switzerland Swiss Data Science Center ETH Zurich EPFL Zurich Switzerland Institute for Machine Learning Department of Computer Science ETH Zurich Switzerland Constructor University Bremen Campus Ring 1 Bremen28759 Germany Innoterra BioScience Private Limited Mumbai India

Implementing image-based machine learning in agriculture is often limited by scarce data and annotations, making it hard to achieve high-quality model predictions. This study tackles the issue of postharvest quality assessment of bananas in decentralized supply chains. We propose a method to detect and segment surface defects in banana images using panoptic segmentation to quantify defect size and number. Instead of time-consuming pixel-level annotations, we use weak supervision with coarse labels. A dataset of 476 smartphone images of bananas was collected under real-world field conditions and annotated for bruises and scars. Using the Segment Anything Model (SAM), a recently published foundation model for image segmentation, we generated dense annotations from coarse bounding boxes to train a segmentation model, significantly reducing manual effort while achieving a panoptic quality score of 77.6%. This demonstrates SAM's potential for low-effort, accurate segmentation in agricultural settings with limited data. © 2024, CC BY.

关键词： Fruits

Your Transformer May Not be as Powerful as You Expect

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Luo, Shengjie Li, Shanda Zheng, Shuxin Liu, Tie-Yan Wang, Liwei He, Di Key Laboratory of Machine Perception MOE School of Intelligence Science and Technology Peking University China Machine Learning Department School of Computer Science Carnegie Mellon University United States Microsoft Research United States Center for Data Science Peking University China Zhejiang Lab China

Relative Positional Encoding (RPE), which encodes the relative distance between any pair of tokens, is one of the most successful modifications to the original Transformer. As far as we know, theoretical understanding of the RPE-based Transformers is largely unexplored. In this work, we mathematically analyze the power of RPE-based Transformers regarding whether the model is capable of approximating any continuous sequence-to-sequence functions. One may naturally assume the answer is in the affirmative-RPE-based Transformers are universal function approximators. However, we present a negative result by showing there exist continuous sequence-to-sequence functions that RPE-based Transformers cannot approximate no matter how deep and wide the neural network is. One key reason lies in that most RPEs are placed in the softmax attention that always generates a right stochastic matrix. This restricts the network from capturing positional information in the RPEs and limits its capacity. To overcome the problem and make the model more powerful, we first present sufficient conditions for RPE-based Transformers to achieve universal function approximation. With the theoretical guidance, we develop a novel attention module, called Universal RPE-based (URPE) Attention, which satisfies the conditions. Therefore, the corresponding URPE-based Transformers become universal function approximators. Extensive experiments covering typical architectures and tasks demonstrate that our model is parameter-efficient and can achieve superior performance to strong baselines in a wide range of applications. The code will be made publicly available at https://***/lsj2408/URPE. Copyright © 2022, The Authors. All rights reserved.

关键词： Stochastic systems

PAM: A Propagation-Based Model for Segmenting Any 3D Objects across Multi-Modal Medical Images

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Chen, Zifan Nan, Xinyu Li, Jiazheng Zhao, Jie Li, Haifeng Lin, Ziling Li, Haoshen Chen, Heyun Liu, Yiting Tang, Lei Zhang, Li Dong, Bin Center for Data Science Peking University Beijing China Department of Radiology Key Laboratory of Carcinogenesis and Translational Research Ministry of Education Peking University Cancer Hospital and Institute Beijing China National Engineering Laboratory for Big Data Analysis and Applications Peking University Beijing China Peking University Beijing China Center for Machine Learning Research Peking University Beijing China National Biomedical Imaging Center Peking University Beijing China

Background: Volumetric segmentation is crucial for medical imaging applications but faces significant challenges. Current approaches often require extensive manual annotations and scenario-specific model training, limiting their transferability across different tasks or modalities. While general segmentation models offer some versatility in natural image processing, they struggle with the unique characteristics of medical images. There is an urgent need in clinical practice for a new segmentation approach that can effectively handle medical imagery features while maintaining adaptability across various three-dimensional objects and imaging modalities. Methods: We introduce PAM (Propagating Anything Model), a propagation-based segmentation approach that operates on 3D medical image volumes using a 2D prompt (bounding box or sketch mask). PAM extrapolates this initial input to generate a complete 3D segmentation by modeling inter-slice structural relationships, establishing a continuous information flow within 3D medical structures. This approach enhances segmentation effectiveness across various imaging modalities by focusing on structural and semantic continuities rather than isolating specific objects. The model combines a CNN-based UNet architecture for intra-slice information processing with a Transformer-based attention module to facilitate inter-slice propagation. This innovative framework results in a method with unique generalizability, capable of segmenting diverse 3D objects across different medical imaging modalities. Results: PAM demonstrated superior performance on 44 diverse medical datasets, notably improving the dice similarity coefficient (DSC) for hundreds of segmentation object types and various medical imaging modalities. Compared to modern models like MedSAM and SegVol, PAM achieved an average DSC improvement of over 18.1%, while maintaining stable predictions despite prompt deviation (one-way ANOVA test, P ≥ 0.5985) and varying propagation confi

关键词： Semantic Segmentation

WiViPose: A Video-aided Wi-Fi Framework for Environment-Independent 3D Human Pose Estimation

学校读者我要写书评

暂无评论

IEEE Transactions on Multimedia 2025年

作者： Zhang, Lei Ning, Haoran Tang, Jiaxin Chen, Zhenxiang Zhong, Yaping Han, Yahong Tianjin University College of Intelligence and Computing the Tianjin Key Laboratory of Advanced Network Technology and Application Tianjin300050 China Key Laboratory of Computing Power Network and Information Security Ministry of Education China University of Jinan Shandong Provincial Key Laboratory of Ubiquitous Intelligent Computing the School of Information Science and Engineering Jinan250022 China Wuhan Sports University Sports Big-data Research Center Wuhan430079 China Tianjin University College of Intelligence and Computing the Tianjin Key Laboratory of Machine Learning Tianjin300350 China

The inherent complexity of Wi-Fi signals makes video-aided Wi-Fi 3D pose estimation difficult. The challenges include the limited generalizability of the task across diverse environments, its significant signal heterogeneity, and its inadequate ability to analyze local and geometric information. To overcome these challenges, we introduce WiViPose, a video-aided Wi-Fi framework for 3D pose estimation, which attains enhanced cross-environment generalization through cross-layer optimization. Bilinear temporal-spectral fusion (BTSF) is initially used to fuse the time-domain and frequency-domain features derived from Wi-Fi. Video features are derived from a multiresolution convolutional pose machine and enhanced by local self-attention. Cross-modality data fusion is facilitated through an attention-based transformer, with the process further refined under a supervisory mechanism. WiViPose demonstrates effectiveness by achieving an average percentage of correct keypoints (PCK)@50 of 91.01% across three typical indoor environments. © 2025 IEEE.

关键词： Frequency domain analysis

Empowering LLMs with Logical Reasoning: A Comprehensive Survey

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Cheng, Fengxiang Li, Haoxuan Liu, Fenrong van Rooij, Robert Zhang, Kun Lin, Zhouchen Institute for Logic Language and Computation University of Amsterdam Netherlands Center for Data Science Peking University China Machine Learning Department MBZUAI Department of Philosophy Tsinghua University China Department of Philosophy CMU United States Institute for Artificial Intelligence Peking University China Peng Cheng Laboratory China National Key Lab of General AI School of Intelligence Science and Technology Peking University China

Large language models (LLMs) have achieved remarkable successes on various natural language tasks. However, recent studies have found that there are still significant challenges to the logical reasoning abilities of LLMs. This paper summarizes and categorizes the main challenges into two aspects: (1) Logical question answering, LLMs often fail to generate the correct answer within complex logical problem which requires sophisticated deductive, inductive or abductive reasoning given a collection of premises and constrains. (2) Logical consistency, LLMs are prone to producing responses contradicting themselves across different questions. For example, a state-of-the-art Macaw question-answering LLM answers Yes to both questions Is a magpie a bird? and Does a bird have wings? but answers No to Does a magpie have wings?. To facilitate this research direction, we comprehensively investigate the most cutting-edge methods and propose detailed taxonomies of these methods. Specifically, to accurately answer complex logic questions, previous methods can be categorized based on reliance on external solvers, prompts, pretraining, and fine-tuning. To avoid logical contradictions, we discuss concepts and solutions of various logical consistencies, including implication, negation, transitivity, factuality consistency, and their composites. In addition, we review commonly used benchmark datasets and evaluation metrics, and discuss promising research directions, such as extensions to modal logic to account for uncertainty, and efficient algorithms satisfying multiple logical consistencies simultaneously. © 2025, CC BY.

关键词： Birds

RVPD: An Automated System for Calculating the Tortuosity and Bifurcation Angles of Retinal Vessels to Predict Diseases

学校读者我要写书评

暂无评论

RVPD: An Automated System for Calculating the Tortuosity and...

IEEE International Symposium on Biomedical Imaging

作者： Guangzhengao Yang Xingyu Luo Jie Zhao Fangfang Fan Haoshen Li Bin Dong Li Zhang Yan Zhang Center for Data Science Peking University China Department of Cardiology Peking University First Hospital China National Engineering Laboratory for Big Data Analysis and Applications Peking University China Peking University Changsha Institute for Computing and Digital Economy China Beijing International Center for Mathematical Research Peking University China Center for Machine Learning Research Peking University China National Biomedical Imaging Center Peking University China

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

Hypertension and diabetes are known to potentially cause morphological changes in the retinal capillary system, yet quantifying these changes presents significant challenges. This research addresses this issue by designing and developing an automated system capable of analyzing retinal images to detect and measure these changes. Our system automatically segments vessels and the optic disc, calculates tortuosity and bifurcation angles, and performs disease prediction based on these measurements along with clinical features. For tortuosity calculation, we adopted a novel curve fitting method, which provides more stable convergence compared to traditional discrete numerical methods. The system integrates clinical data and conducts prediction experiments on hypertensive and diabetic patients with varying sample sizes. Our results demonstrate the system's strong performance and scalability: in the hypertension group, the AUC reached a maximum of 0.814 ± 0.023, with an ACC of 0.735 ± 0.033, while in the diabetes group, the AUC reached a maximum of 0.710 ± 0.043, with an ACC of 0.641 ± 0.052. These findings suggest that our automated system could serve as a valuable tool for early detection and risk assessment of hypertension and diabetes, potentially improving patient outcomes through timely intervention.

关键词： Hypertension Image segmentation Scalability Bifurcation Optical variables measurement Optical imaging Retinal vessels Diabetes Risk management Diseases

Mining Negative Temporal Contexts For False Positive Suppression In Real-Time Ultrasound Lesion Detection

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Yu, Haojun Li, Youcheng Wu, QuanLin Zhao, Ziwei Chen, Dengbo Wang, Dong Wang, Liwei National Key Laboratory of General Artificial Intelligence School of Intelligence Science and Technology Peking University Beijing China Center of Data Science Peking University Beijing China Center for Machine Learning Research Peking University Beijing China Yizhun Medical AI Co. Ltd Beijing China Guangdong China

During ultrasonic scanning processes, real-time lesion detection can assist radiologists in accurate cancer diagnosis. However, this essential task remains challenging and underexplored. General-purpose real-time object detection models can mistakenly report obvious false positives (FPs) when applied to ultrasound videos, potentially misleading junior radiologists. One key issue is their failure to utilize negative symptoms in previous frames, denoted as negative temporal contexts (NTC) [15]. To address this issue, we propose to extract contexts from previous frames, including NTC, with the guidance of inverse optical flow. By aggregating extracted contexts, we endow the model with the ability to suppress FPs by leveraging NTC. We call the resulting model UltraDet. The proposed UltraDet demonstrates significant improvement over previous state-of-the-arts and achieves real-time inference speed. We release the code, checkpoints, and high-quality labels of the CVA-BUS dataset [9] in https://***/HaojunYu1998/UltraDet. © 2023, CC BY-NC-ND.

关键词： Object detection

learning Physics-Informed Neural Networks without Stacked Back-propagation

学校读者我要写书评

暂无评论

arXiv 2022年

作者： He, Di Li, Shanda Shi, Wenlei Gao, Xiaotian Zhang, Jia Bian, Jiang Wang, Liwei Liu, Tie-Yan National Key Laboratory of General Artificial Intelligence School of Intelligence Science and Technology Peking University China Machine Learning Department School of Computer Science Carnegie Mellon University United States Microsoft Research Center for Data Science Peking University China

Physics-Informed Neural Network (PINN) has become a commonly used machine learning approach to solve partial differential equations (PDE). But, facing high-dimensional second-order PDE problems, PINN will suffer from severe scalability issues since its loss includes second-order derivatives, the computational cost of which will grow along with the dimension during stacked back-propagation. In this work, we develop a novel approach that can significantly accelerate the training of Physics-Informed Neural Networks. In particular, we parameterize the PDE solution by the Gaussian smoothed model and show that, derived from Stein's Identity, the second-order derivatives can be efficiently calculated without back-propagation. We further discuss the model capacity and provide variance reduction methods to address key limitations in the derivative estimation. Experimental results show that our proposed method can achieve competitive error compared to standard PINN training but is significantly faster. Our code is released at https://***/LithiumDA/PINN-without-Stacked-BP. Copyright © 2022, The Authors. All rights reserved.

关键词： Backpropagation