检索结果-内蒙古大学图书馆

IDEMPOTENCE AND PERCEPTUAL IMAGE COMPRESSION

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Xu, Tongda Zhu, Ziran He, Dailan Li, Yanghao Guo, Lina Wang, Yuanyuan Wang, Zhe Qin, Hongwei Wang, Yan Liu, Jingjing Zhang, Ya-Qin Institute for AI Industry Research Tsinghua University China Department of Computer Science and Technology Tsinghua University China Institute of Software Chinese Academy of Sciences China SenseTime Research China The Chinese University of Hong Kong Hong Kong School of Vehicle and Mobility Tsinghua University China

Idempotence is the stability of image codec to re-compression. At the first glance, it is unrelated to perceptual image compression. However, we find that theoretically: 1) Conditional generative model-based perceptual codec satisfies idempotence;2) Unconditional generative model with idempotence constraint is equivalent to conditional generative codec. Based on this newfound equivalence, we propose a new paradigm of perceptual image codec by inverting unconditional generative model with idempotence constraints. Our codec is theoretically equivalent to conditional generative codec, and it does not require training new models. Instead, it only requires a pre-trained mean-square-error codec and unconditional generative model. Empirically, we show that our proposed approach outperforms state-of-the-art methods such as HiFiC (Mentzer et al., 2020) and ILLM (Muckley et al., 2023), in terms of Fréchet Inception Distance (FID). The source code is provided in https://***/tongdaxu/ Idempotence-and-Perceptual-Image-Compression. © 2024, CC BY.

关键词： Image coding

Variance-insensitive and Target-preserving Mask Refinement for Interactive Image Segmentation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Fang, Chaowei Zhou, Ziyin Chen, Junye Su, Hanjing Wu, Qingyao Li, Guanbin School of Artificial Intelligence Xidian University Xi’an China School of Computer Science and Engineering Research Institute of Sun Yat-sen University in Shenzhen Sun Yat-sen University Guangzhou China Tencent China School of Software Engineering South China University of Technology Guangzhou China GuangDong Province Key Laboratory of Information Security Technology China

Point-based interactive image segmentation can ease the burden of mask annotation in applications such as semantic segmentation and image editing. However, fully extracting the target mask with limited user inputs remains challenging. We introduce a novel method, Variance-Insensitive and Target-Preserving Mask Refinement to enhance segmentation quality with fewer user inputs. Regarding the last segmentation result as the initial mask, an iterative refinement process is commonly employed to continually enhance the initial mask. Nevertheless, conventional techniques suffer from sensitivity to the variance in the initial mask. To circumvent this problem, our proposed method incorporates a mask matching algorithm for ensuring consistent inferences from different types of initial masks. We also introduce a target-aware zooming algorithm to preserve object information during downsampling, balancing efficiency and accuracy. Experiments on GrabCut, Berkeley, SBD, and DAVIS datasets demonstrate our method’s state-of-the-art performance in interactive image segmentation. Copyright © 2023, The Authors. All rights reserved.

关键词： Semantics

An Efficient GPU-based Implementation for Noise Robust Sound Source Localization

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Lin, Zirui Takigahira, Masayuki Terakado, Naoya Gulzar, Haris Busto, Monikka Roslianna Eda, Takeharu Itoyama, Katsutoshi Nakadai, Kazuhiro Amano, Hideharu Dept. of Systems and Control Engineering School of Engineering Tokyo Institute of Technology Tokyo Japan Honda Research Institute Japan Co. Ltd. Saitama Japan NTT Software Innovation Center Tokyo Japan Dept. of Information and Computer Science Keio University Kanagawa Japan

Robot audition, encompassing Sound Source Localization (SSL), Sound Source Separation (SSS), and Automatic Speech Recognition (ASR), enables robots and smart devices to acquire auditory capabilities similar to human hearing. Despite their wide applicability, processing multi-channel audio signals from microphone arrays in SSL involves computationally intensive matrix operations, which can hinder efficient deployment on Central Processing Units (CPUs), particularly in embedded systems with limited CPU resources. This paper introduces a GPU-based implementation of SSL for robot audition, utilizing the Generalized Singular Value Decomposition-based Multiple Signal Classification (GSVD-MUSIC), a noise-robust algorithm, within the HARK platform, an open-source software suite. For a 60-channel microphone array, the proposed implementation achieves significant performance improvements. On the Jetson AGX Orin, an embedded device powered by an NVIDIA GPU and ARM®Cortex™-A78AE v8.2 64-bit CPUs, we observe speedups of 4645.1× for GSVD calculations and 8.8× for the SSL module, while speedups of 2223.4× for GSVD calculation and 8.95× for the entire SSL module on a server configured with an NVIDIA A100 GPU and AMD EPYC™ 7352 CPUs, making real-time processing feasible for large-scale microphone arrays and providing ample capacity for real-time processing of potential subsequent machine learning or deep learning tasks. © 2025, CC BY-NC-ND.

关键词： Acoustic generators

Limited quantum advantage for stellar interferometry via continuous-variable teleportation

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Huang, Zixin Baragiola, Ben Q. Menicucci, Nicolas C. Wilde, Mark M. School of Mathematical and Physical Sciences Macquarie University NSW2109 Australia Centre for Quantum Software and Information Faculty of Engineering and IT University of Technology Sydney Australia Centre for Quantum Computation and Communication Technology School of Science RMIT University MelbourneVIC3000 Australia School of Electrical and Computer Engineering Cornell University IthacaNY14850 United States

We consider stellar interferometry in the continuous-variable (CV) quantum information formalism and use the quantum Fisher information (QFI) to characterize the performance of three key strategies: direct interferometry (DI), local heterodyne measurement, and a CV teleportation-based strategy. In the lossless regime, we show that a squeezing parameter of r ≈ 2 (18 dB) is required to reach ≈ 95% of the QFI achievable with DI;such a squeezing level is beyond what has been achieved experimentally. In the low-loss regime, the CV teleportation strategy becomes inferior to DI, and the performance gap widens as loss increases. Curiously, in the high-loss regime, a small region of loss exists where the CV teleportation strategy slightly outperforms both DI and local heterodyne, representing a transition in the optimal strategy. We describe this advantage as limited because it occurs for a small region of loss, and the magnitude of the advantage is also small. We argue that practical difficulties further impede achieving any quantum advantage, limiting the merits of a CV teleportation-based strategy for stellar interferometry. Copyright © 2023, The Authors. All rights reserved.

关键词： Interferometry

Machina Economicus: A New Paradigm for Prosumers in the Energy Internet of Smart Cities

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hou, Luyang Yan, Jun Wu, Yuankai Wang, Chun Qiu, Tie School of Computer Science National Pilot Software Engineering School Beijing University of Posts and Telecommunications 10 Xitucheng Rd Haidian District Beijing100876 China Institute for Information Systems Engineering Concordia University 1455 De Maisonneuve Blvd. W. MontréalQCH3G 1M8 Canada College of Computer Science Sichuan University No. 29 Wangjiang Road Wuhou District Sichuan Chengdu61006 China School of Computer Science and Technology College of Intelligence and Computing Tianjin University Tianjin300350 China

Energy Internet (EI) is emerging as new share economy platform for flexible local energy supplies in smart cities. Empowered by the Internet-of-Things (IoT) and Artificial Intelligence (AI), EI aims to unlock peer-to-peer energy trading and sharing among prosumers, who can adeptly switch roles between providers and consumers in localized energy markets with rooftop photovoltaic panels, vehicle-to-everything technologies, packetized energy management, etc. The integration of prosumers in EI, however, will encounter many challenges in modelling, analyzing, and designing an efficient, economic, and social-optimal platform for energy sharing, calling for advanced AI/IoT-based solutions to resource optimization, information exchange, and interaction protocols in the context of the share economy. In this study, we aim to introduce a recently emerged paradigm, Machina Economicus, to investigate the economic rationality in modelling, analysis, and optimization of AI/IoT-based EI prosumer behaviors. The new paradigm, built upon the theory of machine learning and mechanism design, will offer new angles to investigate the selfishness of AI through a game-theoretic perspective, revealing potential competition and collaborations resulting from the self-adaptive learning and decision-making capacity. This study will focus on how the introduction of AI will reshape prosumer behaviors on the EI, and how this paradigm will reveal new research questions and directions when AI meets the share economy. With an extensive case analysis in the literature, we will also shed light on potential solutions for advancements of AI in future smart cities. Copyright © 2024, The Authors. All rights reserved.

关键词： Internet of things

A Novel Meta-Hierarchical Active Inference Reinforcement Learning Approach for QoS-Driven Resource Allocation in Dynamic Clouds

学校读者我要写书评

暂无评论

A Novel Meta-Hierarchical Active Inference Reinforcement Lea...

IEEE Conference on Global Communications (GLOBECOM)

作者： Peijie Xian Ying He F. Richard Yu Jianbo Du College of Computer Science and Software Engineering Shenzhen Univ. Shenzhen China Shaanxi Key Lab. of Information Comm. Network and Security Xi’an Univ. of Posts & Telecomm. Xi’an China School of Information Technology Carleton Univ. Ottawa Canada

ISBN: (数字)9798350351255

ISBN: (纸本)9798350351262

The cloud computing environment is highly dynamic due to a variety of external factors such as seasonal changes, market trends, and social events. In this case, tenant behavior patterns exhibit significant variability. Therefore, the cloud computing resource allocation model must continuously adapt to these changes. To this end, we propose an adaptive algorithm for dynamic cloud resource allocation based on meta-hierarchical active inference reinforcement learning (MHAIRL). The algorithm combines active inference with meta-hierarchical reinforcement learning. It improves the overall performance of the algorithm, as well as quickly adapts to environmental changes, and improves the generalization performance. In addition, we design a novel polling scheduling framework combined with long short-term memory (LSTM) network. The framework ensures scheduling fairness and flexibility while greatly reducing the state and action space dimensions of the agent. Extensive simulation results show that our method outperforms baseline algorithms in quality of service (QoS) metrics and significantly improves system performance in highly dynamic cloud resource allocation.

关键词： Cloud computing Adaptation models Heuristic algorithms Computational modeling Quality of service Reinforcement learning Dynamic scheduling Inference algorithms Resource management Long short term memory

Multi-View Representation Learning for Multi-Instance Learning with Applications to Medical Image Classification

学校读者我要写书评

暂无评论

Multi-View Representation Learning for Multi-Instance Learni...

IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

作者： Lu Zhao Liming Yuan Zhenliang Li Xianbin Wen School of Computer and Information Engineering Tianjin Chengjian University Tianjin China School of Computer Science and Engineering Tianjin University of Technology Tianjin China Tianjin Key Laboratory of Intelligence Computing and Novel Software Technology Tianjin China Key Laboratory of Computer Vision and System Ministry of Education Tianjin China

ISBN: (数字)9781665468190

ISBN: (纸本)9781665468206

Multi-Instance Learning (MIL) is a weakly supervised learning paradigm, in which every training example is a labeled bag of unlabeled instances. In typical MIL applications, instances are often used for describing the features of regions/parts in a whole object, e.g., regional patches/lesions in an eye-fundus image. However, for a (semantically) complex part the standard MIL formulation puts a heavy burden on the representation ability of the corresponding instance. To alleviate this pressure, we still adopt a bag-of-instances as an example in this paper, but extract from each instance a set of representations using $1 \times1$ convolutions. The advantages of this tactic are two-fold: i) This set of representations can be regarded as multi-view representations for an instance; ii) Compared to building multi-view representations directly from scratch, extracting them automatically using $1 \times1$ convolutions is more economical, and may be more effective since $1 \times1$ convolutions can be embedded into the whole network. Furthermore, we apply two consecutive multi-instance pooling operations on the reconstituted bag that has actually become a bag of sets of multi-view representations. We have conducted extensive experiments on several canonical MIL data sets from different application domains. The experimental results show that the proposed framework outperforms the standard MIL formulation in terms of classification performance and has good interpretability.

关键词： Training Representation learning Retinopathy Supervised learning Buildings Diabetes Bioinformatics

STDEN: Towards Physics-Guided Neural Networks for Traffic Flow Prediction

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Ji, Jiahao Wang, Jingyuan Jiang, Zhe Jiang, Jiawei Zhang, Hu State Key Laboratory of Software Development Environment School of Computer Science & Engineering Beihang University Beijing China Peng Cheng Laboratory Shenzhen China Department of Computer & Information Science & Engineering The University of Florida United States MOE Engineering Research Center of Advanced Computer Application Technology School of Computer Science & Engineering Beihang University Beijing China

High-performance traffic flow prediction model designing, a core technology of Intelligent Transportation System, is a long-standing but still challenging task for industrial and academic communities. The lack of integration between physical principles and data-driven models is an important reason for limiting the development of this field. In the literature, physics-based methods can usually provide a clear interpretation of the dynamic process of traffic flow systems but are with limited accuracy, while data-driven methods, especially deep learning with black-box structures, can achieve improved performance but can not be fully trusted due to lack of a reasonable physical basis. To bridge the gap between purely data-driven and physics-driven approaches, we propose a physics-guided deep learning model named Spatio- Temporal Differential Equation Network (STDEN), which casts the physical mechanism of traffic flow dynamics into a deep neural network framework. Specifically, we assume the traffic flow on road networks is driven by a latent potential energy field (like water flows are driven by the gravity field), and model the spatio-temporal dynamic process of the potential energy field as a differential equation network. STDEN absorbs both the performance advantage of data-driven models and the interpretability of physics-based models, so is named a physics-guided prediction model. Experiments on three real-world traffic datasets in Beijing show that our model outperforms state-of-the-art baselines by a significant margin. A case study further verifies that STDEN can capture the mechanism of urban traffic and generate accurate predictions with physical meaning. The proposed framework of differential equation network modeling may also cast light on other similar applications. Copyright © 2022, The Authors. All rights reserved.

关键词： Differential equations

Hierarchical software Design Methodology for High Concurrency and Fine-Grained Permission Control Scenarios

学校读者我要写书评

暂无评论

Hierarchical Software Design Methodology for High Concurrenc...

2021年第二届控制、机器人和智能系统国际会议

作者： Yuan Han Qingqing Song Xinquan Huang Jie Yin School of Computer and Software Nanjing Vocational University of Industry Technology School of electronic & information engineering Suzhou University of Science and Technology

Because complex software usually faces high concurrency and complex permission control scenarios,and there are problems with low code reuse rate and chaotic function module structure in the development process of its access control module,so providing software design methodology for such scenarios is helpful to improve software productivity and software quality play an important *** paper proposes a hierarchical software design methodology for the development of complex software under high concurrency and fine-grained permission control *** software design methodology divides the hierarchical structure according to the characteristics of the access control process,and provides support for fine-grained control and multi-level control of *** high concurrency scenarios,a cache management layer is added to improve the efficiency of the access control *** to the actual needs of software development,the specific levels in the software design methodology are *** decoupling the user status detection,user status verification and permission authentication in the access control process,the support of this software design methodology in different framework environments is guaranteed and the security risks caused by the high coupling between frameworks are *** a series of 0-500 K concurrent test experiments,the performance data of different frameworks in the same scenario is obtained,which provides users with a reference when implementing the software design methodology proposed in this article,enable it to choose a framework for better performance depending on the scenario it faces.

关键词： software Design Methodology High Concurrency Fine-Grained Access Control