检索结果-内蒙古大学图书馆

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

Instance-aware Dynamic Prompt Tuning for Pre-trained Point Cloud Models

学校读者我要写书评

暂无评论

Instance-aware Dynamic Prompt Tuning for Pre-trained Point C...

International Conference on computer Vision (ICCV)

作者： Yaohua Zha Jinpeng Wang Tao Dai Bin Chen Zhi Wang Shu-Tao Xia Tsinghua Shenzhen International Graduate School Tsinghua University College of Computer Science and Software Engineering Shenzhen University Harbin Institute of Technology Shenzhen Research Center of Artificial Intelligence Peng Cheng Laboratory

Pre-trained point cloud models have found extensive applications in 3D understanding tasks like object classification and part segmentation. However, the prevailing strategy of full fine-tuning in downstream tasks leads to large per-task storage overhead for model parameters, which limits the efficiency when applying large-scale pre-trained models. Inspired by the recent success of visual prompt tuning (VPT), this paper attempts to explore prompt tuning on pre-trained point cloud models, to pursue an elegant balance between performance and parameter efficiency. We find while instance-agnostic static prompting, e.g. VPT, shows some efficacy in downstream transfer, it is vulnerable to the distribution diversity caused by various types of noises in real-world point cloud data. To conquer this limitation, we propose a novel Instance-aware Dynamic Prompt Tuning (IDPT) strategy for pre-trained point cloud models. The essence of IDPT is to develop a dynamic prompt generation module to perceive semantic prior features of each point cloud instance and generate adaptive prompt tokens to enhance the model's robustness. Notably, extensive experiments demonstrate that IDPT outperforms full finetuning in most tasks with a mere 7% of the trainable parameters, providing a promising solution to parameter-efficient learning for pre-trained point cloud models. Code is available at https://***/zyh16143998882/ICCV23-IDPT.

关键词：

New Insights on Relieving Task-Recency Bias for Online Class Incremental Learning

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Liang, Guoqiang Chen, Zhaojie Chen, Zhaoqiang Ji, Shiyu Zhang, Yanning Northwestern Polytechnical University 710072 China National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology China Key Laboratory of Big Data Storage and Management China School of Computer Science China School of Software Engineering China

—To imitate the ability of keeping learning of human, continual learning which can learn from a never-ending data stream has attracted more interests recently. In all settings, the online class incremental learning (OCIL), where incoming samples from data stream can be used only once, is more challenging and can be encountered more frequently in real world. Actually, all continual learning models face a stability-plasticity dilemma, where the stability means the ability to preserve old knowledge while the plasticity denotes the ability to incorporate new knowledge. Although replay-based methods have shown exceptional promise, most of them concentrate on the strategy for updating and retrieving memory to keep stability at the expense of plasticity. To strike a preferable trade-off between stability and plasticity, we propose an Adaptive Focus Shifting algorithm (AFS), which dynamically adjusts focus to ambiguous samples and non-target logits in model learning. Through a deep analysis of the task-recency bias caused by class imbalance, we propose a revised focal loss to mainly keep stability. By utilizing a new weight function, the revised focal loss will pay more attention to current ambiguous samples, which are the potentially valuable samples to make model progress quickly. To promote plasticity, we introduce a virtual knowledge distillation. By designing a virtual teacher, it assigns more attention to non-target classes, which can surmount overconfidence and encourage model to focus on inter-class information. Extensive experiments on three popular datasets for OCIL have shown the effectiveness of AFS. The code will be available at https://***/czjghost/AFS. Copyright © 2023, The Authors. All rights reserved.

关键词： Distillation

MOKD: Cross-domain Finetuning for Few-shot Classification via Maximizing Optimized Kernel Dependence

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Tian, Hongduan Liu, Feng Liu, Tongliang Du, Bo Cheung, Yiu-Ming Han, Bo TMLR Group Hong Kong Baptist University Hong Kong Department of Computer Science Hong Kong Baptist University Hong Kong TMLR Group University of Melbourne Australia Sydney AI Centre The University of Sydney Australia National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence School of Computer Science Wuhan Univeristy China

In cross-domain few-shot classification, nearest centroid classifier (NCC) aims to learn representations to construct a metric space where few-shot classification can be performed by measuring the similarities between samples and the prototype of each class. An intuition behind NCC is that each sample is pulled closer to the class centroid it belongs to while pushed away from those of other classes. However, in this paper, we find that there exist high similarities between NCC-learned representations of two samples from different classes. In order to address this problem, we propose a bi-level optimization framework, maximizing optimized kernel dependence (MOKD) to learn a set of class-specific representations that match the cluster structures indicated by labeled data of the given task. Specifically, MOKD first optimizes the kernel adopted in Hilbert-Schmidt independence criterion (HSIC) to obtain the optimized kernel HSIC (opt-HSIC) that can capture the dependence more precisely. Then, an optimization problem regarding the opt-HSIC is addressed to simultaneously maximize the dependence between representations and labels and minimize the dependence among all samples. Extensive experiments on Meta-Dataset demonstrate that MOKD can not only achieve better generalization performance on unseen domains in most cases but also learn better data representation clusters. The project repository of MOKD is available at: https://***/tmlr-group/MOKD. Copyright © 2024, The Authors. All rights reserved.

关键词： Machine learning

Discrete RKZD (Reciprocal-Kind Zhang Dynamics) Algorithm Solving Time-Independent Linear Matrix-Vector Equation (TILMVE) Simplified from Time-Dependent Situation

学校读者我要写书评

暂无评论

Discrete RKZD (Reciprocal-Kind Zhang Dynamics) Algorithm Sol...

Chinese Control Conference (CCC)

作者： Yunong Zhang Ji Lu Mingzhi Mao Shuai Li School of Computer Science and Engineering Sun Yat-sen University Guangzhou P. R. China School of Software Engineering Sun Yat-sen University Zhuhai P. R. China Faculty of Information Technology and Electrical Engineering University of Oulu Oulu Finland

We present a continuous reciprocal-kind Zhang dynamics (RKZD) model for solving the time-dependent linear matrixvector equation. On the basis of the model, we deduce its simplified form for solving the time-independent linear matrix-vector equation (TILMVE). Subsequently, for more efficient computation and easier implementation in digital hardware, we utilize Euler forward difference formula (EFDF) to discretize the continuous RKZD model, resulting in a discrete RKZD algorithm. Finally, numerical experimental results attest to the feasibility and high effectiveness of the discrete RKZD algorithm for solving TILMVE. Comparisons with the discrete gradient neural network (or termed discrete gradient dynamics), Jacobi iteration, as well as Gauss-Seidel iteration highlight the superior convergence properties of the discrete RKZD algorithm.

关键词：

Hidden high-risky states identification from routine urban traffic

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Liu, Shiyan Bai, Mingyang Guo, Shengmin Gao, Jianxi Sun, Huijun Gao, Ziyou Li, Daqing School of Reliability and Systems Engineering Beihang University Beijing100191 China State Key Laboratory of Software Development Environment Beihang University Beijing100191 China Department of Computer Science Rensselaer Polytechnic Institute TroyNY12180 United States School of Systems Science Beijing Jiaotong University Beijing100044 China

One of the core risk management tasks is to identify hidden high-risky states that may lead to system breakdown, which can provide valuable early warning knowledge. However, due to high dimensionality and nonlinear interaction embedded in large-scale complex systems like urban traffic, it remains challenging to identify hidden high-risky states from huge system state space where over 99% of possible system states are not yet visited in empirical data. Based on maximum entropy model, we infer the underlying interaction network from complicated dynamical processes of urban traffic, and construct system energy landscape. In this way, we can locate hidden high-risky states that have never been observed from real data. These states can serve as risk signals with high probability of entering hazardous minima in energy landscape, which lead to huge recovery cost. Our finding might provide insights for complex system risk management. Significance Statement Hidden high-risky states are lethal system states with catastrophic consequences for large-scale complex systems. Existing methods for identifying hidden high-risky states face challenges in routine urban traffic, where over 99% of possible system states are not yet visited in empirical data. Based on large-scale traffic-speed data, we build a least-structured network entropy model that captures intricate traffic interactions. This model can generate the entire state space and help the identification of hidden high-risky states that may lead to extreme events. Our results suggest a new way to provide reliable early warning signals for system breakdown based on critical risk scenarios. Copyright © 2024, The Authors. All rights reserved.

关键词： Large scale systems

MANDARIN LOMBARD FLAVOR CLASSIFICATION

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Liu, Qingmu Yang, Yuhong Li, Baifeng Chen, Hongyang Tu, Weiping Lin, Song National Engineering Research Center for Multimedia Software School of Computer Science Wuhan University Wuhan China Hubei Key Laboratory of Multimedia and Network Communication Engineering Wuhan China Guangdong OPPO Mobile Telecommunications Corp. China

The Lombard effect refers to individuals’ unconscious modulation of vocal effort in response to variations in the ambient noise levels, intending to enhance speech intelligibility. The impact of different decibel levels and types of background noise on Lombard effects remains unclear. Building upon the characteristic of Lombard speech that individuals adjust their speech to improve intelligibility dynamically based on the self-feedback speech, we propose a flavor classification approach for the Lombard effect. We first collected Mandarin Lombard speech under different noise conditions, then simulated self-feedback speech, and ultimately conducted the statistical test on the word correct rate. We found that both SSN and babble noise types result in four distinct categories of Mandarin Lombard speech in the range of 30 to 80 dBA with different transition points. Copyright © 2023, The Authors. All rights reserved.

关键词： Speech intelligibility

Action-Manipulation Attack and Defense to X-Armed Bandits

学校读者我要写书评

暂无评论

Action-Manipulation Attack and Defense to X-Armed Bandits

IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom)

作者： Zhi Luo Youqi Li Lixing Chen Zichuan Xu Pan Zhou School of Cyber Science and Engineering Huazhong University of Science and Technology School of Computer Science School of Cyberspace Science and Technology Beijing Institute of Technology Institute of Cyber Science and Technology Shanghai Jiao Tong University School of Software Dalian University of Technology Hubei Engineering Research Center on Big Data Security School of Cyber Science and Engineering Huazhong University of Science and Technology

As a continuous variant of Multi-armed bandits (MAB), $\mathcal{X}$-armed bandits have enriched many applications of online machine learning like personalized recommendation system. However, the attack and defense to the $\mathcal{X}$-armed bandits remain largely unexplored, though the MAB has proved to be vulnerable. In this paper, we aim to bridge this gap and investigate the robustness analysis for the $\mathcal{X}$-armed bandits. Specifically, we consider action-manipulation attack, which is practical but harder than the existing reward-manipulation attack. We propose an attack algorithm based on a lower bound tree (LBT), which can continuously hijack the learner’s action by perturbing $\mathcal{X}$-armed bandits’ high confidence tree (HCT) construction. As a result, the nodes including the arm targeted by the attacker is selected frequently with a sublinear attack cost. To defend against the LBT attack, we propose a robust version of the HCT algorithm, called RoHCT. We theoretically analyze that the regret of RoHCT is related to the upper bound of the total cost Q and still sublinear to total number of rounds T. We carry out experiments to evaluate the effectiveness of LBT and RoHCT.

关键词： Threat modeling Bridges Privacy Costs Upper bound Machine learning Big Data

FineDiffusion: Scaling up Diffusion Models for Fine-grained Image Generation with 10,000 Classes

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Pan, Ziying Wang, Kun Li, Gang He, Feihong Lai, Yongxuan School of Informatics Xiamen University China Institute of Software Chinese Academy of Sciences China School of Mathematics and Information Engineering Longyan University China Shenyang Institute of Computing Technology of Chinese Academy of Sciences China University of Chinese Academy of Sciences China School of Computer Science and Technology Soochow University China

The class-conditional image generation based on diffusion models is renowned for generating high-quality and diverse images. However, most prior efforts focus on generating images for general categories, e.g., 1000 classes in ImageNet-1k. A more challenging task, large-scale fine-grained image generation, remains the boundary to explore. In this work, we present a parameter-efficient strategy, called FineDiffusion, to fine-tune large pre-trained diffusion models scaling to large-scale fine-grained image generation with 10,000 categories. FineDiffusion significantly accelerates training and reduces storage overhead by only fine-tuning tiered class embedder, bias terms, and normalization layers' parameters. To further improve the image generation quality of fine-grained categories, we propose a novel sampling method for fine-grained image generation, which utilizes superclass-conditioned guidance, specifically tailored for fine-grained categories, to replace the conventional classifier-free guidance sampling. Compared to full fine-tuning, FineDiffusion achieves a remarkable 1.56× training speed-up and requires storing merely 1.77% of the total model parameters, while achieving state-of-the-art FID of 9.776 on image generation of 10,000 classes. Extensive qualitative and quantitative experiments demonstrate the superiority of our method compared to other parameter-efficient fine-tuning methods. The code and more generated results are available at our project website: https://***/. © 2024, CC BY.

关键词： Image enhancement