检索结果-内蒙古大学图书馆

Interference-Aware Opportunistic Job Placement for Shared Distributed Deep Learning Clusters

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Li, Hongliang Zhao, Hairui Sun, Ting Li, Xiang Xu, Haixiao Li, Keqin College of Computer Science and Technology Jilin University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering The Ministry of Education Changchun China High Performance Computing Center Jilin University Changchun China Department of Computer Science State University of New York New PaltzNY12561 United States

Distributed deep learning frameworks facilitate large deep learning workloads. These frameworks support sharing one GPU device among multiple jobs to improve resource utilization. Modern deep learning training jobs consume a large amount of GPU memory. Despite that, sharing GPU memory among jobs is still possible because a training job has iterative steps that its memory usage fluctuates over time. However, resource sharing also introduces the risk of job performance degradation. Colocated jobs sharing a GPU device may suffer from different levels of interference, mainly caused by memory oversharing. How to improve resource utilization while maintaining good job performance is a novel challenge for job placement strategies. This paper studies the job placement problem. We propose an opportunistic memory sharing model to describe the time-varying job memory requirements. Based on this model, we introduce an Opportunistic Job Placement Problem (OJPP) for shared GPU clusters that seek job placement configurations using a minimum number of GPU devices and guarantee user-defined performance requirements at the same time. We propose a greedy algorithm and a heuristic algorithm with computational complexities of O(n log n) and O(n2 log n), respectively, to solve the problem. We also propose an online adjustment algorithm with the computational complexity of O(n log n) to perform updates to job placement configurations in runtime. A machine-learning-based interference prediction method is used to prepare accurate interference estimations. Extensive experiments are conducted on a GPU cluster to verify the correctness and effectiveness of our algorithms. Compared with standalone training jobs on dedicated clusters, the proposed approach reduces resource consumption by 46% in a shared cluster, while guaranteeing over 92.97% of the job performance, in terms of average job completion time. © 2023, The Authors. All rights reserved.

关键词： computational complexity

Weakly supervised prototype topic model with discriminative seed words: Modifying the category prior by self-exploring supervised signals

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wang, Bing Wang, Yue Li, Ximing Ouyang, Jihong College of Computer Science and Technology Jilin University China Key Laboratory of Symbolic Computation Knowledge Engineering of Ministry of Education Jilin University China

Dataless text classification, i.e., a new paradigm of weakly supervised learning, refers to the task of learning with unlabeled documents and a few predefined representative words of categories, known as seed words. The recent generative dataless methods construct document-specific category priors by using seed word occurrences only, however, such category priors often contain very limited and even noisy supervised signals. To remedy this problem, in this paper we propose a novel formulation of category prior. First, for each document, we consider its label membership degree by not only counting seed word occurrences, but also using a novel prototype scheme, which captures pseudo-nearest neighboring categories. Second, for each label, we consider its frequency prior knowledge of the corpus, which is also a discriminative knowledge for classification. By incorporating the proposed category prior into the previous generative dataless method, we suggest a novel generative dataless method, namely Weakly Supervised Prototype Topic Model (Wsptm). The experimental results on real-world datasets demonstrate that Wsptm outperforms the existing baseline methods. © 2021, CC0.

关键词： Text processing

CellMirror: Deciphering Cell Populations from Spatial Transcriptomics Data by Interpretable Contrastive Learning 1

学校读者我要写书评

暂无评论

CellMirror: Deciphering Cell Populations from Spatial Transc...

1st IEEE International Conference on Medical Artificial Intelligence, MedAI 2023

作者： Xia, Junjie Cui, Jiangnan Huang, Zhenyu Zhang, Shuangquan Yao, Fang Zhang, Yijian Zuo, Chunman Institute of Artificial Intelligence Donghua University Department of Applied Mathematics Shanghai China College of Computer Science and Technology Jilin University Changchun China School of Cyber Science and Engineering Nanjing University of Science and Technology Nanjing China Institute of Maternal and Child Medicine Southern Medical University Affiliated Shenzhen Maternity and Child Healthcare Hospital Shenzhen China Xinhua Hospital Affiliated Shanghai Jiao Tong University of Medicine Department of General Surgery Shanghai China Institute of Artificial Institute Donghua University Shanghai China Jilin University Key Laboratory of Symbolic Computation and Knowledge Engineering Ministry of Education Changchun China

ISBN: (纸本)9798350358780

Spatial transcriptomics (ST) has enabled us elucidating tumor microenvironments, however, the technological limitations without single-cell resolution severely hinder its application. Here, we propose CellMirror, an interpretable contrastive learning model to decipher heterogeneous cell populations in ST data by single-cell RNA-sequencing (scRNA-seq) data. Specifically, CellMirror learns the disentangled shared (representing biological variations in both data) and salient features (specific to ST data) by two contrastive variational encoders, while constructing the relations between genes and features by a shared linear decoder. In various cancer samples, CellMirror outperforms other tools in learning common features for label transfer, and interpretation of features. Particularly, in breast cancer studies, CellMirror detects finer domains in ST data missed by other methods, and is robust to dissect cell populations in ST data using independent scRNA-seq data. These results demonstrate applications of CellMirror in interpreting the complex tumor structure in ST data by integrating scRNA-seq data. © 2023 IEEE.

关键词： Data integration

BARGAIN-MATCH: A Game Theoretical Approach for Resource Allocation and Task Offloading in Vehicular Edge Computing Networks

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Sun, Zemin Sun, Geng Liu, Yanheng Wang, Jian Cao, Dongpu The College of Computer Science and Technology Jilin University Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun130012 China The School of Vehicle and Mobility Tsinghua University Beijing China

Vehicular edge computing (VEC) is emerging as a promising architecture of vehicular networks (VNs) by deploying the cloud computing resources at the edge of the VNs. However, efficient resource management and task offloading in the VEC network is challenging. In this work, we first present a hierarchical framework that coordinates the heterogeneity among tasks and servers to improve the resource utilization for servers and service satisfaction for vehicles. Moreover, we formulate a joint resource allocation and task offloading problem (JRATOP), aiming to jointly optimize the intra-VEC server resource allocation and inter-VEC server load-balanced offloading by stimulating the horizontal and vertical collaboration among vehicles, VEC servers, and cloud server. Since the formulated JRATOP is NP-hard, we propose a cooperative resource allocation and task offloading algorithm named BARGAIN-MATCH, which consists of a bargaining-based incentive approach for intra-server resource allocation and a matching method-based horizontal-vertical collaboration approach for inter-server task offloading. Besides, BARGAIN-MATCH is proved to be stable, weak Pareto optimal, and polynomial complex. Simulation results demonstrate that the proposed approach achieves superior system utility and efficiency compared to the other methods, especially when the system workload is heavy. Copyright © 2022, The Authors. All rights reserved.

关键词： Game theory

Joint Power and 3D Trajectory Optimization for UAV-enabled Wireless Powered Communication Networks with Obstacles

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Pan, Hongyang Liu, Yanheng Sun, Geng Fan, Junsong Liang, Shuang Yuen, Chau The College of Computer Science and Technology Jilin University Changchun130012 China Pillar Singapore University of Technology and Design 487372 Singapore The Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin University Changchun130012 China The School of Information Science and Technology Northeast Normal University Changchun130024 China

Unmanned aerial vehicle (UAV)-enabled wireless powered communication networks (WPCNs) are promising technologies in 5G/6G wireless communications, while there are several challenges about UAV power allocation and scheduling to enhance the energy utilization efficiency, considering the existence of obstacles. In this work, we consider a UAV-enabled WPCN scenario that a UAV needs to cover the ground wireless devices (WDs). During the coverage process, the UAV needs to collect data from the WDs and charge them simultaneously. To this end, we formulate a joint-UAV power and three-dimensional (3D) trajectory optimization problem (JUPTTOP) to simultaneously increase the total number of the covered WDs, increase the time efficiency, and reduce the total flying distance of UAV so as to improve the energy utilization efficiency in the network. Due to the difficulties and complexities, we decompose it into two sub optimization problems, which are the UAV power allocation optimization problem (UPAOP) and UAV 3D trajectory optimization problem (UTTOP), respectively. Then, we propose an improved non-dominated sorting genetic algorithm-II with K-means initialization operator and Variable dimension mechanism (NSGA-II-KV) for solving the UPAOP. For UTTOP, we first introduce a pretreatment method, and then use an improved particle swarm optimization with Normal distribution initialization, Genetic mechanism, Differential mechanism and Pursuit operator (PSO-NGDP) to deal with this sub optimization problem. Simulation results verify the effectiveness of the proposed strategies under different scales and settings of the networks. Copyright © 2023, The Authors. All rights reserved.

关键词： Energy utilization

A Multi-Agent Deep Reinforcement Learning Method for Fully Noisy Observations

学校读者我要写书评

暂无评论

SSRN

SSRN 2023年

作者： Wang, Kaiyu Zhang, Menglin Qu, Bohao Wang, Xianchang College of Computer Science and Technology Jilin University Jilin Changchun130012 China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Jilin Changchun130012 China School of Artificial Intelligence Jilin University Jilin Changchun130012 China Chengdu Kestrel Artificial Intelligence Institute Chengdu611730 China

Multi-agent reinforcement learning (MARL) algorithms have achieved great breakthroughs in many aspects. The MARL algorithms can learn effective policies in ideal simulation environments. But different from the ideal simulation environments, noise is unavoidable in the real world. MARL algorithms need to learn effective policies in unavoidable fully noisy environments. In this paper, we consider a challenging multi-agent reinforcement learning problem: All agents cannot observe any noiseless observations from environments during the whole training process and MARL algorithms cannot learn effective policies in these fully noisy observation environments. To solve this problem, we propose a method called Robust Policy Learning under Fully Noisy Observation viA DeNoising REpresentation NeTwork (PLANET), which enables MARL algorithms learning effective policies in fully noisy observation environments. The PLANET method learns the effective policy through two steps. (1) Extracting the noise characteristics and motion laws information from fully noisy observation histories. (2) Making MARL algorithms extract information from the noise characteristics and motion laws information and learn effective policies. The results of a series of exhaustive experiments show that our method can eliminate the effects of noise and learn effective policies in fully noisy observation environments. © 2023, The Authors. All rights reserved.

关键词： Learning algorithms

Neural Combinatorial Optimization Algorithms for Solving Vehicle Routing Problems: A Comprehensive Survey with Perspectives

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wu, Xuan Wang, Di Wen, Lijie Xiao, Yubin Wu, Chunguo Wu, Yuesong Yu, Chaoyu Maskell, Douglas L. Zhou, You Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education College of Computer Science and Technology Jilin university Changchun130012 China Joint NTU-UBC Research Centre of Excellence in Active Living for the Elderly Nanyang Technological University 639798 Singapore School of Software Tsinghua University Beijing100084 China College of Computing and Data Science Nanyang Technological University 639798 Singapore

Although several surveys on Neural Combinatorial Optimization (NCO) solvers specifically designed to solve Vehicle Routing Problems (VRPs) have been conducted, they did not cover the state-of-the-art (SOTA) NCO solvers emerged recently. More importantly, to establish a comprehensive and up-to-date taxonomy of NCO solvers, we systematically review relevant publications and preprints, categorizing them into four distinct types, namely Learning to Construct, Learning to Improve, Learning to Predict-Once, and Learning to Predict-Multiplicity solvers. Subsequently, we present the inadequacies of the SOTA solvers, including poor generalization, incapability to solve large-scale VRPs, inability to address most types of VRP variants simultaneously, and difficulty in comparing these NCO solvers with the conventional Operations Research algorithms. Simultaneously, we discuss on-going efforts, identify open inadequacies, as well as propose promising and viable directions to overcome these inadequacies. Notably, existing efforts focus on only one or two of these inadequacies, with none attempting to address all of them concurrently. In addition, we compare the performance of representative NCO solvers from the Reinforcement, Supervised, and Unsupervised Learning paradigms across VRPs of varying scales. Finally, following the proposed taxonomy, we provide an accompanying web page as a live repository for NCO solvers. Through this survey and the live repository, we aim to foster further advancements in the NCO community. Copyright © 2024, The Authors. All rights reserved.

关键词： Combinatorial optimization

Categorizing npm Packages by Analyzing the Text Information in Software Repositories

学校读者我要写书评

暂无评论

Categorizing npm Packages by Analyzing the Text Information ...

Asia-Pacific Conference on Software engineering

作者： Yu Wang Huaxiao Liu Shanquan Gao Shujia Li College of computer science and technology Jilin University Changchun China Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education Changchun China

ISBN: (纸本)9781665437851

To prevent JavaScript developers from reinventing wheels, npm ecosystem provides numerous third-party libraries for developers to realize relevant functionalities. Npm displays the tags provided by the creators for these packages to help developers find suitable ones. However, not all creators have the habit of tagging their packages, and thus npm cannot provide tag information of a lot of packages for developers to help them understand the package functionalities effectively. Considering that many tags are unrelated to the functionality of packages, we propose a method to find out the tags that are important to distinguish the functionality categories of packages and assign them to untagged packages for assisting developers in the process of retrieving the packages. Firstly, we analyze the attribute of existing tags in npm to establish category tags (functionality categories). Then, we further mine the readme of tagged packages to generate keywords for each category tag. Finally, our method identifies category tags for untagged packages by measuring the similarity between their readme and the keywords of category tags. The evaluation demonstrates that our approach has a good performance in assigning category tags to untagged packages.

关键词： Correlation Codes Ecosystems Wheels Tagging Software Libraries

Robust Misinformation Detection by Visiting Potential Commonsense Conflict

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Wang, Bing Li, Ximing Li, Changchun Zhao, Bingrui Fu, Bo Guan, Renchu Wang, Shengsheng College of Computer Science and Technology Jilin University China Key Laboratory of Symbolic Computation and Knowledge Engineering of the MoE Jilin University China School of Computer Science and Artificial Intelligence Liaoning Normal University China

The development of Internet technology has led to an increased prevalence of misinformation, causing severe negative effects across diverse domains. To mitigate this challenge, Misinformation Detection (MD), aiming to detect online misinformation automatically, emerges as a rapidly growing research topic in the community. In this paper, we propose a novel plug-and-play augmentation method for the MD task, namely Misinformation Detection with Potential Commonsense Conflict (MD-PCC). We take inspiration from the prior studies indicating that fake articles are more likely to involve commonsense conflict. Accordingly, we construct commonsense expressions for articles, serving to express potential commonsense conflicts inferred by the difference between extracted commonsense triplet and golden ones inferred by the well-established commonsense reasoning tool COMET. These expressions are then specified for each article as augmentation. Any specific MD methods can be then trained on those commonsense-augmented articles. Besides, we also collect a novel commonsense-oriented dataset named CoMis, whose all fake articles are caused by commonsense conflict. We integrate MD-PCC with various existing MD backbones and compare them across 4 public benchmark datasets and CoMis. Empirical results demonstrate that MD-PCC can consistently outperform the existing MD baselines. Copyright © 2025, The Authors. All rights reserved.

关键词：