检索结果-内蒙古大学图书馆

SSRN 2023年

作者： Yang, Haoran Wang, Junli Duan, Rui Yan, Chungang Key Laboratory of Embedded System and Service Computing Ministry of Education Shanghai201804 China Collaborative Innovation Center for Financial Network Security Tongji University Shanghai201804 China

Deep clustering plays an important role in data analysis, and with the prevalence of graph data nowadays, various deep clustering models on graph are constantly proposed. However, due to the lack of more adequate clustering guidance, the discriminability of feature representation learned from these models for the clustering task is limited. Therefore, for the purpose of enabling the output of these models to be more cluster-oriented, we propose a Deep Clustering Optimization Method for Graph Neural networks (DCOM-GNN), which can be attached to the original model architecture conveniently. For DCOM-GNN, it contains two components, one is the inter-cluster distance optimization module, whose role is to further adjust the distance between clusters of the original model output rationally. Another one is the intra-cluster distance optimization module, which aims to improve the cohesiveness of the original model output. Comprehensive experiments show that the performance of various deep clustering models on graph can be significantly improved after adding DCOM-GNN. © 2023, The Authors. All rights reserved.

关键词： Cluster analysis

来源：评论

学校读者我要写书评

暂无评论

Graph Ensemble Neural network

SSRN

引用

SSRN 2023年

作者： Duan, Rui Yan, Chungang Wang, Junli Jiang, Changjun Key Laboratory of Embedded System and Service Computing Ministry of Education Shanghai201804 China Collaborative Innovation Center for Financial Network Security Tongji University Shanghai201804 China

Ensemble methods have been shown to improve graph neural networks (GNNs). Existing ensemble methods on graphs determine a strong classifier by combining a set of trained base classifiers, i.e., combining the final outputs of base classifiers for prediction. However, these methods fail to promote many popular GNNs to perform well under heterophily (in graphs where many connected nodes havedifferent class labels), which limits their applicability. Furthermore, they ignore the hierarchical nature of GNNs, which results in no interaction between base classifiers when neighbors are aggregated (during training). Two issues arise from this: low applicability and shallow ensemble. We propose Graph Ensemble Neural network (GEN) for addressing above issues, which is not a simple ensembleof GNNs, but instead integrates ensemble into GNNs to fuse a set of graphs. GEN deepens single ensemble into multiple ensembles during training and applies to homophily and heterophily graphs. In GEN, we design structure augmentation to generate some graphs for training and design feature augmentation for attenuating errors brought by the initial features. Different from existing graphensemble methods that execute only one ensemble, GEN executes multiple deep ensembles throughout the neighbor aggregation to fuse multiple graphs generated by structure augmentation. Extensive experiments show that GEN achieves new state-of-the-art performance on homophily and heterophily graphs for the semi- and full-supervised node classification. The source code of GEN is publicly available at https://***/graphNN/GEN1. © 2023, The Authors. All rights reserved.

关键词： Graph neural networks

来源：评论

学校读者我要写书评

暂无评论

A Novel ICD Coding Method Based on Associated and Hierarchical Code Description Distillation

arXiv

引用

arXiv 2024年

作者： Zhang, Bin Wang, Junli Key Laboratory of Embedded System and Service Computing Tongji University Ministry of Education Shanghai201804 China Collaborative Innovation Center for Financial Network Security Tongji University Shanghai201804 China

ICD(International Classification of Diseases) coding involves assigning ICD codes to patients visit based on their medical notes. ICD coding is a challenging multilabel text classification problem due to noisy medical document inputs. Recent advancements in automated ICD coding have enhanced performance by integrating additional data and knowledge bases with the encoding of medical notes and codes. However, most of them ignore the code hierarchy, leading to improper code assignments. To address these problems, we propose a novel method based on associated and hierarchical code description distillation (AHDD) for better code representation learning and avoidance of improper code *** utilize the code description and the hierarchical structure inherent to the ICD codes. Therefore, in this paper, we leverage the code description and the hierarchical structure inherent to the ICD codes. The code description is also applied to aware the attention layer and output layer. Experimental results on the benchmark dataset show the superiority of the proposed method over several state-of-the-art baselines. Copyright © 2024, The Authors. All rights reserved.

关键词： Benchmarking

来源：评论

学校读者我要写书评

暂无评论

Routing Strategy for SDN Large Flow Based on Deep Reinforcement Learning

Routing Strategy for SDN Large Flow Based on Deep Reinforcem...

引用

IEEE International Conference on Big Data and Cloud computing (BdCloud)

作者： Yu Ke Junli Wang Chungang Yan Jiamin Yao Key Laboratory of Embedded System and Service Computing Collaborative Innovation Center for Financial Network Security (Tongji University) Ministry of Education National (Province-Ministry Joint) Tongji University Shanghai China

With the expansion of network scale, traditional routing strategies in Software-Defined networking (SDN) have difficulty adapting to fast traffic variability. The large flow in the network occupies 80% of the network traffic. An unreasonable schedule of the large flow will lead to the emergence of bottleneck links and reduce the network performance. To solve the above problems, this paper proposes a different routing method, called Routing Strategy for SDN Large Flow Based on Deep Reinforcement Learning (RSDRL). Considering the links in the network that can carry large flow are limited, RSDRL can schedule large flow firstly and generate routing strategies adaptively. At the same time, in order to avoid the occurrence of congestion, RSDRL quantifies the degree of congestion in the network to implement routing. Experiments show that the proposed model has better Quality of Service (QoS) performance under the test traffic, realizing a reasonable schedule of traffic.

关键词： Deep learning Training Knowledge engineering Schedules Adaptation models Reinforcement learning Quality of service

来源：评论

学校读者我要写书评

暂无评论

Correct after Answer: Enhancing Multi-Span Question Answering with Post-Processing Method

arXiv

引用

arXiv 2024年

作者： Lin, Jiayi Zhang, Chenyang Tong, Haibo Zhang, Dongyu Hong, Qingqing Hou, Bingxuan Wang, Junli Key Laboratory of Embedded System and Service Computing Tongji University Ministry of Education Shanghai201804 China Collaborative Innovation Center for Financial Network Security Tongji University Shanghai201804 China

Multi-Span Question Answering (MSQA) requires models to extract one or multiple answer spans from a given context to answer a question. Prior work mainly focuses on designing specific methods or applying heuristic strategies to encourage models to predict more correct predictions. However, these models are trained on gold answers and fail to consider the incorrect predictions. Through a statistical analysis, we observe that models with stronger abilities do not predict less incorrect predictions compared with other models. In this work, we propose Answering-Classifying-Correcting (ACC) framework, which employs a post-processing strategy to handle incorrect predictions. Specifically, the ACC framework first introduces a classifier to classify the predictions into three types and exclude "wrong predictions", then introduces a corrector to modify "partially correct predictions". Experiments on several MSQA datasets show that ACC framework significantly improves the Exact Match (EM) scores, and further analysis demostrates that ACC framework efficiently reduces the number of incorrect predictions, improving the quality of predictions.1 © 2024, CC BY-NC-ND.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

FedDDB: Clustered Federated Learning based on Data Distribution Difference 22

FedDDB: Clustered Federated Learning based on Data Distribut...

引用

Proceedings of the 2022 5th International Conference on Algorithms, computing and Artificial Intelligence

作者： Chengyu You Zihao Lu Junli Wang Chungang Yan Key Laboratory of Embedded System and Service Computing (Tongji University) Ministry of Education China and National (Province-Ministry Joint) Collaborative Innovation Center for Financial Network Security Tongji University China

ISBN: (纸本)9781450398336

Clustered federated learning is a federated learning method based on multi-task learning. It groups similar clients into the same clusters and shares model parameters to solve the problem that the joint model is trapped in local optima on Non-IID data. Most of the existing clustered federated learning methods are based on the difference of model parameters for clients clustering. During the client model training process, the model parameters are biased and the clustering result is affected due to insufficient samples and missing eigenvalues in the dataset. In this paper, we develop a clustered federated learning method based on data distribution difference (FedDDB) in the dataset level. The method in this paper focuses on the distribution of label probability and eigenvalues, analyzes the difference of data distribution difference between clients and measures the distance between datasets which is used for client clustering. Every cluster will be trained independently and in parallel on the cluster center model. At the beginning of each round of training, the client clustering process needs to be repeated. We conduct relevant experiments and demonstrate the effectiveness of our method.

关键词： Clustering method

来源：评论

学校读者我要写书评

暂无评论

A Dual-stage Attention Based SDN Traffic Prediction Method

A Dual-stage Attention Based SDN Traffic Prediction Method

引用

IEEE International Conference on Big Data and Cloud computing (BdCloud)

作者： Chenjing Gao Junli Wang Chungang Yan Key Laboratory of Embedded System and Service Computing (Tongji University) Ministry of Education National (Province-Ministry Joint) Collaborative Innovation Center for Financial Network Security Tongji University Shanghai China

Traffic matrix is the main research object of traffic prediction in software-defined networking. Accurate and timely traffic matrix prediction plays an important role in avoiding network congestion. While various methods have been proposed in previous studies to solve the SDN traffic prediction problem, few of them consider both inter- and intra-flow features. In this paper, we propose a dual-stage attention based Traffic Prediction method for the SDN traffic matrix prediction task. First, our method uses temporal pattern attention as the inter-flow attention mechanism before the encoding stage to realize adaptive feature extraction. Then, temporal attention is introduced as the intra-flow attention mechanism before the decoding stage to capture long-term temporal dependencies. Finally, we use the autoregressive module to handle the highly dynamic SDN traffic volume. Experimental results show that our proposed SDN traffic prediction method can capture more traffic features and improve the SDN traffic matrix prediction performance to some extent.

关键词： network topology Time series analysis Prediction methods Machine learning Feature extraction Graph neural networks Encoding

来源：评论

学校读者我要写书评

暂无评论

LAMPT: LAbel Mask-Predicted Transformer for Extreme Multi-label Text Classification

LAMPT: LAbel Mask-Predicted Transformer for Extreme Multi-la...

引用

IEEE International Conference on Big Data

作者： Yifan Xu Minghao Zhu Junli Wang Chungang Yan Key Laboratory of Embedded System and Service Computing (Tongji University) Ministry of Education. National (Province-Ministry Joint) Collaborative Innovation Center for Financial Network Security Tongji University Shanghai China

ISBN: (纸本)9781665480468

Extreme Multi-label text Classification ( XMC) is a task of recalling the most relevant labels for each given text from an extremely large-scale label set. It is emphasized that XMC is a more complex classification task because there are two main problems: large space of labels and the labels in XMC tasks tend to be correlated. Existing methods attempt to model label correlations by viewing the XMC task as a sequence generation problem however they still suffer from (1) using the slow serial decoding strategy where labels are predicted one-by-one. (2) needing to compare a mass of label ordering strategies in the decoding stage to achieve satisfied accuracy. In this work, we propose LAbel Mask-Predicted Transformer (LAMPT) to address the both issues, which is a novel non-autoregressive generation model that (1) enriches the input raw text representation with the additional label features by fully exploiting the label dependencies, (2) allows for efficient parallel decoding thanks to its non-autoregressive decoding formulation and mask-prediced training strategy. Experimental results demonstrate that single model performance is substantially enhanced by LAMPT. On a Wiki dataset with thirty-one thousand labels, LAMPT-XLNet accuracy has gained 1.6% relative improvement on P@3 over the LightXML-XLNet. Also, the P@1 of ensemble LAMPT is 90.00%, a significant i mprovement over the state-of-the-art ensemble LightXML (transformer-based) and AttentionXML (LSTM-based), which achieve 89.45% and 87.47%, respectively.

关键词： Training Correlation Text categorization Information services Big Data Predictive models Transformers

来源：评论

学校读者我要写书评

暂无评论

An Energy-efficient Traffic Scheduling Method based on Slime Mould Algorithm for SDN

An Energy-efficient Traffic Scheduling Method based on Slime...

引用

International Conference on Information Systems and Computer networks (ISCON)

作者： Zheyuan Wang Junli Wang Chungang Yan Ministry of Education Key Laboratory of Embedded System and Service Computing (Tongji University) Shanghai China National (Province-Ministry Joint) Collaborative Innovation Center for Financial Network Security Tongji University Shanghai China

Software defined network (SDN) enables efficient and green traffic management by separating the control and data planes. However, the existing itemized scheduling approach is prone to waste of network resources and energy consumption due to the different sensitivity of traffic to time delay. To address these problems, we design a Slime Mould Algorithm based Energy-efficient Traffic Scheduling Method (SMA-ETSM) for SDN. First, we model the traffic scheduling problem with delay requirements to evaluate the decision. Second, in order to minimize the generation of invalid solutions in traffic scheduling, the encoding of the slime mould algorithm is improved, and a slime mould adaptation and update mechanism suitable for energy-efficient traffic scheduling is designed. The experiments show that SMA-ETSM can effectively reduce the energy consumption and improve the overall bandwidth utilization of the network compared with ECMP and ACO algorithm, and it also has some improvement in the operation efficiency compared with the original slime mould algorithm.

关键词： Energy consumption Sensitivity Processor scheduling Scheduling Energy efficiency Real-time systems Time factors

来源：评论

学校读者我要写书评

暂无评论

An Improved Convolutional Neural network Algorithm for Multi-Label Classification

An Improved Convolutional Neural Network Algorithm for Multi...

引用

International Conference on Audio, Language and Image Processing, ICALIP

作者： Xinsheng Wang Lijun Sun Zhihua Wei Department of Computer Science and Technology Tongji University Shanghai China Research Center of Big Data and Network Security Tongji University Shanghai China Key Laboratory of Embedded System and Service Computing Tongji University Shanghai China

Recent years conventional neural network(CNN) has been applied to different natural language processing(NLP) tasks such as sentence classification, sentence modeling, etc. Some researchers use CNN to do multi-label classification but their work mainly focus on image rather than text. In this paper, we propose an improved CNN via hierarchical dirichlet process(HDP) model to deal with the multi-label classification problem in NLP. We first apply an HDP model to discard some words which are less important semantically. Then we use word embedding methods to transform words to vectors. Finally, we train CNN based on word vectors. Experimental results demonstrate that our method is superior to most traditional multi-label classification methods and TextCNN in terms of performance.

关键词： Feature extraction Training Convolutional neural networks Natural language processing Vocabulary Transforms

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：