检索结果-内蒙古大学图书馆

Prevent the language model from being overconfident in neural machine translation

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Miao, Mengqi Meng, Fandong Liu, Yijin Zhou, Xiao-Hua Zhou, Jie Peking University China Pattern Recognition Center WeChat AI Tencent Inc China Beijing International Center for Mathematical Research National Engineering Lab for Big Data Analysis and Applications Department of Biostatistics Peking University Beijing China

The Neural Machine Translation (NMT) model is essentially a joint language model conditioned on both the source sentence and partial translation. Therefore, the NMT model naturally involves the mechanism of the Language Model (LM) that predicts the next token only based on partial translation. Despite its success, NMT still suffers from the hallucination problem, generating fluent but inadequate translations. The main reason is that NMT pays excessive attention to the partial translation while neglecting the source sentence to some extent, namely overconfidence of the LM. Accordingly, we define the Margin between the NMT and the LM, calculated by subtracting the predicted probability of the LM from that of the NMT model for each token. The Margin is negatively correlated to the overconfidence degree of the LM. Based on the property, we propose a Margin-based Token-level Objective (MTO) and a Margin-based Sentencelevel Objective (MSO) to maximize the Margin for preventing the LM from being overconfident. Experiments on WMT14 English-to-German, WMT19 Chinese-to-English, and WMT14 English-to-French translation tasks demonstrate the effectiveness of our approach, with 1.36, 1.50, and 0.63 BLEU improvements, respectively, compared to the Transformer baseline. The human evaluation further verifies that our approaches improve translation adequacy as well as fluency. Copyright © 2021, The Authors. All rights reserved.

关键词： Neural machine translation

Electrochemical Parameter Identification for Lithium-ion Battery Sources in Self-Sustained Transportation Energy Systems

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Gu, Yuxuan Wang, Jianxiao Chen, Yuanbo Zheng, Kedi Deng, Zhongwei Chen, Qixin The Department of Electrical Engineering Tsinghua University Beijing100084 China The National Engineering Laboratory for Big Data Analysis and Applications Peking University Beijing100871 China The College of Mechanical and Vehicle Engineering Chongqing University Chongqing400044 China

Lithium-ion battery (LIB) sources have played an essential role in self-sustained transportation energy systems and have been widely deployed in the last few years. To realize reliable battery maintenance, identifying its electrochemical parameters is necessary. However, the battery model contains many parameters while the measurable states are only the current and voltage, inducing the identification inherently an ill-conditioned problem. A parameter identification approach is proposed, including the experiment, model, and algorithm. Electrochemical parameters are first grouped manually based on the physical properties and assigned to two sequenced tests for identification. The two tests named the quasi-static test and the dynamic test, are compressed on time for practical implementation. Proper optimization models and a sensitivity-oriented stepwise (SSO) optimization algorithm are developed to search for the optimal parameters efficiently. Typically, the Sobol method is applied to conduct the sensitivity analysis. Based on the sensitivity indexes, the SSO algorithm can decouple the mixed impacts of different parameters during the identification. For validation, numerical experiments on a typical NCM811 battery at different life stages are conducted. The proposed approach saves about half the time finding the proper parameter value. The identification accuracy of crucial parameters related to battery degradation can exceed 95%. Case study results indicate that the identified parameters can not only improve the accuracy of the battery model but also be used as the indicator of the battery SOH. Copyright © 2022, The Authors. All rights reserved.

关键词： Lithium-ion batteries

Cascading Fault Prevention of Power Grid Based on Key Power Generation Nodes 1

学校读者我要写书评

暂无评论

16th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2020 in conjunction with the 13th International Conference on Frontiers of Information Technology, applications and Tools, FITAT 2020

作者： Chang, Kuo-Chi Luo, Jie Deng, Hui-Qiong Li, Qin-Bin Zheng, Rong-Jin Li, Pei-Qiang School of Information Science and Engineering Fujian University of Technology Fuzhou350108 China Fujian Provincial University Engineering Research Genter of Smart Grid Simulation Analysis and Integrated Control Fuzhou350108 China Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology Fuzhou China College of Mechanical and Electrical Engineering National Taipei University of Technology Taipei Taiwan Department of Business Administration North Borneo University College Sabah Malaysia

ISBN: (数字)9789813364202

ISBN: (纸本)9789813364196

In the case of power grid cascading trip, this study puts forward a strategy to prevent cascading trip by changing the output of generator set. Firstly, according to the specific performance of the power grid under the cascade tripping, combined with the action behavior of the line current-type backup protection, and an expression equation for measuring the safety level index between the power grid and the cascading trip failure is given. Secondly, the vulnerable branch is found out through the safety level index, the output of the key power generation nodes on the branch is taken as the optimization variable, and the optimization model is established to improve the safety level index under the chain trip of the power grid. Finally, a combination of sensitivity analysis and particle swarm optimization(PSO) was used to analyze an example on the IEEE-39 node system. The results demonstrate the feasibility and effectiveness of the model and method, and provide a reference for preventing chain trip accidents. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Sensitivity analysis

Battery Valuation and Management for Battery Swapping Station with an Intertemporal Framework

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Chen, Xinjiang Yang, Yu Wang, Jianxiao Song, Jie He, Guannan College of Engineering Peking University Beijing100091 China Department of Industrial and Systems Engineering University of Florida GainesvilleFL116595 United States National Engineering Laboratory for Big Data Analysis and Applications Peking University Beijing100091 China

Battery swapping as a business model for battery energy storage (BES) has great potential in future integrated low-carbon energy and transportation systems. However, frequent battery swapping will inevitably accelerate battery degradation and shorten the battery life accordingly. To model the tradeoff of BES use between energy and transportation applications coupled by battery swapping, we develop a life-cycle decision model that coordinates battery charging and swapping. This model is derived based on an improved intertemporal decision framework, in which the optimal marginal degradation cost (MDC) of BES is determined to maximize the BES benefit across time and application. The proposed framework and model are applied to manage a battery swapping station that simultaneously provides battery swapping services to electric vehicle customers and provides flexibility service to the power grid, including energy arbitrage and reserve. The case study shows that while the end of the physical life of BES occurs faster with battery swapping, the economic life becomes considerably longer. The results also reveal that the optimal MDC depends on the battery values in each application, and we analyze how the battery swapping price affects the optimal MDC and battery life. The proposed framework and model can also provide decision support for on-demand BES service, such as battery trading, renting and secondary use. Copyright © 2023, The Authors. All rights reserved.

关键词： Decision support systems

Optimal Power Generation Output Considering Cascading Failure 1

学校读者我要写书评

暂无评论

作者： Deng, Hui-Qiong Luo, Jie Li, Qin-Bin Zheng, Rong-Jin Li, Pei-Qiang Chang, Kuo-Chi School of Information Science and Engineering Fujian University of Technology Fuzhou350108 China Fujian Provincial University Engineering Research Genter of Smart Grid Simulation Analysis and Integrated Control Fuzhou350108 China Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of Technology Fuzhou China College of Mechanical and Electrical Engineering National Taipei University of Technology Taipei Taiwan Department of Business Administration North Borneo University College Sabah Malaysia

ISBN: (数字)9789813364202

ISBN: (纸本)9789813364196

Aiming at the cascading fault phenomenon of power system, this paper analyzes the disturbed branches involved in the initial fault and their relationship with the power generating nodes. Firstly, according to the performance of the initial fault branch in the cascading fault phenomenon and the action of relay protection, an expression form of evaluating the severity of the initial fault branch is given. Secondly, the relationship between the current of the residual system branch and the injection power of the nodes after the initial fault removal is analyzed. Then, on the basis of synthesizing the severity of the disturbed branch and its correlation with the power generating nodes, the key power generating nodes on the seriously disturbed branch are selected by clustering analysis. Finally, based on the key power generation nodes, the optimal power generation output model considering the power generation operation cost was established, the model was solved by particle swarm optimization (PSO), the model was simulated by IEEE39 bus system in MATlab environment, and the correctness of the established model and method was verified. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Particle swarm optimization (PSO)

BIM: Improving Graph Neural Networks with Balanced Influence Maximization 40

学校读者我要写书评

暂无评论

BIM: Improving Graph Neural Networks with Balanced Influence...

40th IEEE International Conference on data engineering, ICDE 2024

作者： Zhang, Wentao Gao, Xinyi Yang, Ling Cao, Meng Huang, Ping Shan, Jiulong Yin, Hongzhi Cui, Bin Peking University Center for Machine Learning Research China Institute of Advanced Algorithms Research Shanghai China National Engineering Labratory for Big Data Analytics and Applications The University of Queensland Australia Peking University Key Lab of High Confidence Software Technologies China Apple Inc. Institute of Computational Social Science Peking University Qingdao China

ISBN: (纸本)9798350317152

The imbalanced data classification problem has aroused lots of concerns from both academia and industry since data imbalance is a widespread phenomenon in many real-world scenarios. Although this problem has been well researched from the view of imbalanced class samples, we further argue that graph neural networks (GNNs) expose a unique source of imbalance from the influenced nodes of different classes of labeled nodes, i.e., labeled nodes are imbalanced in terms of the number of nodes they influenced during the influence propagation in GNNs. To tackle this previously unexplored influence-imbalance issue, we connect social influence maximization with the imbalanced node classification problem and propose balanced influence maximization (BIM). Specifically, BIM greedily assigns the pseudo label to the node which can maximize the number of influenced nodes in GNN training while making the influence of each class more balance. Experimental results on five public datasets demonstrate the effectiveness of our method in relieving the influence-imbalance issue. For example, when training a GCN with an imbalance ratio of 0.1, BIM significantly outperforms the most competitive baseline by 0.6% -9.8% in five public datasets in terms of the F1 score. © 2024 IEEE.

关键词： Graph neural networks

Graph Attention Multi-Layer Perceptron

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Zhang, Wentao Yin, Ziqi Sheng, Zeang Li, Yang Ouyang, Wen Li, Xiaosen Tao, Yangyu Yang, Zhi Cui, Bin School of CS & Key Laboratory of High Confidence Software Technologies Peking University China Center for Data Science Peking University National Engineering Laboratory for Big Data Analysis and Applications China Tencent Inc China Beijing Institute of Technology China

Graph neural networks (GNNs) have achieved great success in many graph-based applications. However, the enormous size and high sparsity level of graphs hinder their applications under industrial scenarios. Although some scalable GNNs are proposed for large-scale graphs, they adopt a fixed k-hop neighborhood for each node, thus facing the over-smoothing issue when adopting large propagation depths for nodes within sparse regions. To tackle the above issue, we propose a new GNN architecture — Graph Attention Multi-Layer Perceptron (GAMLP), which can capture the underlying correlations between different scales of graph knowledge. We have deployed GAMLP in Tencent with the Angel platform 1, and we further evaluate GAMLP on both real-world datasets and large-scale industrial datasets. Extensive experiments on these 14 graph datasets demonstrate that GAMLP achieves state-of-the-art performance while enjoying high scalability and efficiency. Specifically, it outperforms GAT by 1.3% regarding predictive accuracy on our large-scale Tencent Video dataset while achieving up to 50× training speedup. Besides, it ranks top-1 on both the leader-boards of the largest homogeneous and heterogeneous graph (i.e., ogbn-papers100M and ogbn-mag) of Open Graph Benchmark 2 Copyright © 2022, The Authors. All rights reserved.

关键词： Graph neural networks

NC-ALG: Graph-Based Active Learning under Noisy Crowd 40

学校读者我要写书评

暂无评论

NC-ALG: Graph-Based Active Learning under Noisy Crowd

40th IEEE International Conference on data engineering, ICDE 2024

作者： Zhang, Wentao Wang, Yexin You, Zhenbang Li, Yang Cao, Gang Yang, Zhi Cui, Bin Center for Machine Learning Research Peking University China Key Lab of High Confidence Software Technologies Peking University China Institute of Advanced Algorithms Research Shanghai China Institute of Computational Social Science Peking University Qingdao China National Engineering Labratory for Big Data Analytics and Applications China TEG Tencent Inc. Department of Data Platform China Beijing Academy of Artificial Intelligence China

ISBN: (纸本)9798350317152

Graph Neural Networks (GNNs) have achieved great success in various data mining tasks but they heavily rely on a large number of annotated nodes, requiring considerable human efforts. Despite the effectiveness of existing GNN-based Active Learning (AL) methods, they assume that the annotated labels are always correct, which is contradictory to the error-prone labeling process in a practical crowdsourcing environment. Besides, due to this impractical assumption, existing works only focus on optimizing the node selection in AL but neglect optimizing the labeling process. Therefore, we present NC-ALG, the first GNN-based AL framework that optimizes both the node selection and node labeling process under a noisy crowd. For node selection, NC-ALG introduces a new measurement to model influence reliability and an effective influence maximization objective to select nodes. For node labeling, NC-ALG significantly reduces the labeling cost by considering the model-predicted labels and the labels of mirror nodes. To the best of our knowledge, this is the first attempt to consider GNN-based AL under the practical noisy crowd. Empirical studies on public datasets demonstrate that NC-ALG significantly outperforms existing methods in terms labeling efficiency. Notably, it only takes NC-ALG one-third of the labeling budget that the competitive baseline GRAIN needs to achieve an accuracy of 70.7 % on PubMed. © 2024 IEEE.

关键词： Crowdsourcing

Finding Simplex Items in data Streams 39

学校读者我要写书评

暂无评论

Finding Simplex Items in Data Streams

39th IEEE International Conference on data engineering, ICDE 2023

作者： Fan, Zhuochen Guo, Jiarui Li, Xiaodong Yang, Tong Zhao, Yikai Wu, Yuhan Cui, Bin Xu, Yanwei Uhlig, Steve Zhang, Gong Peking University School of Computer Science National Engineering Laboratory for Big Data Analysis Technology and Application Beijing China Peng Cheng Laboratory Shenzhen China Central Research Institute 2012 Labs Huawei Technologies Co. Ltd. Theory Lab Hong Kong Queen Mary University of London School of Electronic Engineering and Computer Science London United Kingdom

ISBN: (纸本)9798350322279

In this paper, we propose a new type of item in data streams, called simplex items. Simplex items have frequencies in consecutive p windows that can be approximated by a polynomial of degree at most k, where k = 0, 1, 2. These low-order representable simplex items have a wide range of potential applications. For example, when k = 1, we can leverage these items whose frequency has obvious linear increase or decrease to speed up the running time of a class of machine learning models and detect network attacks such as distributed denial-of-service (DDoS), etc. To find k-degree simplex items in real time, we propose a novel sketch, namely X-Sketch, to accurately record simplex items in a compact space. The key idea of X-Sketch is to effectively filter out non-simplex items with less memory overhead, and then monitor the remaining potential simplex items and keep those items with more consecutive windows. We conduct extensive experiments, and the experimental results show that the F1 Score of X-Sketch is on average 68.6%, 57.9%, and 42.2% higher than the baseline solution for k = 0, 1, 2, respectively. Finally, we also provide a case study that applies X-Sketch to "accelerate"the two machine learning models through end-to-end experiments. We have released our source code at GitHub. © 2023 IEEE.

关键词： Denial-of-service attack