检索结果-内蒙古大学图书馆

Learning to Branch in Combinatorial Optimization With Graph Pointer Networks

IEEE/CAA Journal of Automatica Sinica 2024年第1期11卷 157-169页

作者： Rui Wang Zhiming Zhou Kaiwen Li Tao Zhang Ling Wang Xin Xu Xiangke Liao IEEE the Xiangjiang Laboratory and the College of Systems Engineering National University of Defense TechnologyChangsha 410073China the Institute of Automation Chinese Academy of SciencesBeijing 100190China the College of Systems Engineering National University of Defense TechnologyChangsha 410073 the Hunan Key Laboratory of Multi-Energy System Intelligent Interconnection Technology Changsha 410073China the Department of Automation Tsinghua UniversityBeijing 100084China the College of Intelligence Science and Technology National University of Defense TechnologyChangsha 410073China the College of Computer Science and Technology National University of Defense TechnologyChangsha 410073China

Traditional expert-designed branching rules in branch-and-bound(B&B) are static, often failing to adapt to diverse and evolving problem instances. Crafting these rules is labor-intensive, and may not scale well with complex *** the frequent need to solve varied combinatorial optimization problems, leveraging statistical learning to auto-tune B&B algorithms for specific problem classes becomes attractive. This paper proposes a graph pointer network model to learn the branch rules. Graph features, global features and historical features are designated to represent the solver state. The graph neural network processes graph features, while the pointer mechanism assimilates the global and historical features to finally determine the variable on which to branch. The model is trained to imitate the expert strong branching rule by a tailored top-k Kullback-Leibler divergence loss function. Experiments on a series of benchmark problems demonstrate that the proposed approach significantly outperforms the widely used expert-designed branching rules. It also outperforms state-of-the-art machine-learning-based branch-and-bound methods in terms of solving speed and search tree size on all the test instances. In addition, the model can generalize to unseen instances and scale to larger instances.

关键词： Branch-and-bound(B&B) combinatorial optimization deep learning graph neural network imitation learning

来源：评论

学校读者我要写书评

暂无评论

Enhancing Student Dropout Prediction Using Chi-Square, SMOTE-ENN, and Hyperparameter Tuning of Random Forest

Enhancing Student Dropout Prediction Using Chi-Square, SMOTE...

引用

International Conference on Informatics and Computing (ICIC)

作者： Andri Roni Yunis Djoni Ng Poi Wong Robin Darwin Program of Computer Science Department Universitas Mikroskil Medan Indonesia Program of Information Systems Department Universitas Mikroskil Medan Indonesia Program of Computer Science Department University of Pelita Harapan Tangerang Indonesia

ISBN: (数字)9798331517601

ISBN: (纸本)9798331517618

Reducing dropout rates is crucial for enhancing human capital and education standards. Existing methods, such as Random Forest with Chi-Square and SMOTE-ENN, effectively addressed class imbalance and improved prediction accuracy for dropout data. However, there is still a research gap in achieving optimal model performance. This study addresses the gap by incorporating hyperparameter tuning alongside ChiSquare for feature selection and SMOTE-ENN for handling class imbalance. The dataset was segmented into training and evaluation subsets through the implementation of 10 -fold cross-validation. The testing was conducted with seven variations, namely building and implementing a Random Forest model using the default parameters from the Weka tool and applying six different hyperparameter tuning techniques. The results showed that Hyperband, along with other techniques like TPE, RandomSearch, and BO-TPE, led to substantial improvements in model accuracy, precision, and F-measure, and achieved perfect AUC scores. However, BO-GP and Nevergrad did not improve model performance. These findings suggest that the combination of SMOTE-ENN, Chi-Square, and hyperparameter tuning can enhance the effectiveness of dropout prediction models, with potentially positive implications for early intervention strategies in educational institutions.

关键词： Training Accuracy Training data Predictive models Feature extraction Reliability Tuning Random forests Standards Testing

来源：评论

学校读者我要写书评

暂无评论

Deep Learning-Driven Graft Survival Prediction for Indian AFLD Patients Following Liver Transplantation

Deep Learning-Driven Graft Survival Prediction for Indian A...

引用

Congress on Smart Computing Technologies, CSCT 2023

作者： Rajeev, Devi Remya, S. Nayyar, Anand Nair, Krishnanunni School of Computing Amrita Vishwa Vidyapeetham Amritapuri Kollam India IoT and Intelligent Systems Lab School of Computer Science Duy Tan University Danang Viet Nam Department of Gastrointestinal Surgery School of Medicine Kochi India

ISBN: (纸本)9789819750801

Alcoholic Fatty Liver Disease (AFLD) is a condition in India that sometimes leads to the need for liver transplantation in severe cases. However, accurately predicting the success of liver transplantation Patients with AFLD can be challenging due to the country’s distinct genetic composition, lifestyle differences, and healthcare characteristics. To resolve this problem, we provide a deep learning-based technique that considers categorical and numerical information specifically focused on India patient population. Analyzing a data set of 1000 cases and using a multi-modal deep learning algorithm, our proposed model achieved an accuracy of 95%, successfully identifying the specific risk factors unique to India. This research also focused on the importance of data privacy. Our AI-driven healthcare solution shows promise in diagnosing the various health issues related to liver diseases. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Alcoholic fatty liver disease Deep learning Graft survival predict Healthcare Indian specific risk factor Liver transplantation Multi-model fusion SHAP techniques

来源：评论

学校读者我要写书评

暂无评论

Chat with MES: LLM-driven user interface for manipulating garment manufacturing system through natural language

引用

Journal of Manufacturing systems 2025年 80卷 1093-1107页

作者： Yuan, Zhaolin Li, Ming Liu, Chang Han, Fangyuan Huang, Haolun Dai, Hong-Ning Institute of Artificial Intelligence University of Science and Technology Beijing Beijing China Shunde Innovation School University of Science and Technology Beijing Guangdong China Department of Industrial and Systems Engineering The Hong Kong Polytechnic University Hong Kong Research Institute of Advanced Manufacturing The Hong Kong Polytechnic University Hong Kong Research Centre for Digital Transformation of Tourism The Hong Kong Polytechnic University Hung Hom Hong Kong Centre for Transformative Garment Production Hong Kong Science Park Pak Shek Kok Hong Kong Department of Computer Science Hong Kong Baptist University Hong Kong School of Intelligent Systems Science and Engineering Jinan University Zhuhai China

This paper presents Chat with MES (CWM), an AI agent system, which integrates LLMs into the Manufacturing Execution System (MES), serving as the "ears, mouth, and the brain". This system promotes a paradigm shift in MES interactions from Graphical User Interface (GUI) to natural language interface", offering a more natural and efficient way for workers to manipulate the manufacturing system. Compared with the traditional GUI, both the maintenance costs for developers and the learning costs and the complexity of use for workers are significantly reduced. This paper also contributes two technical improvements to address the challenges of using LLM-Agent in serious manufacturing scenarios. The first one is Request Rewriting, designed to rephrase or automatically follow up on non-standardized and ambiguous requests from users. The second innovation is the Multi-Step Dynamic Operations Generation, which is a pre-execution planning technique similar to Chain-of-Thought (COT), used to enhance the success rate of handling complex tasks involving multiple operations. A case study conducted on a simulated garment MES with 55 manually designed requests demonstrates the high execution accuracy of CWM (80%) and the improvement achieved through query rewriting (9.1%) and Multi-Step Dynamic operations generation (18.2%). The source code of CWM, along with the simulated MES and benchmark requests, is publicly accessible. © 2025 The Authors

关键词： Graphical user interfaces

来源：评论

学校读者我要写书评

暂无评论

UNSUPERVISED SEMANTIC SEGMENTATION WITH SELF-SUPERVISED OBJECT-CENTRIC REPRESENTATIONS 11

UNSUPERVISED SEMANTIC SEGMENTATION WITH SELF-SUPERVISED OBJE...

引用

11th International Conference on Learning Representations, ICLR 2023

作者： Zadaianchuk, Andrii Kleindessner, Matthaeus Zhu, Yi Locatello, Francesco Brox, Thomas Max-Planck Institute for Intelligent Systems Tübingen Germany Amazon Web Services United States Department of Computer Science ETH Zürich Switzerland University of Freiburg Germany

In this paper, we show that recent advances in self-supervised representation learning enable unsupervised object discovery and semantic segmentation with a performance that matches the state of the field on supervised semantic segmentation 10 years ago. We propose a methodology based on unsupervised saliency masks and self-supervised feature clustering to kickstart object discovery followed by training a semantic segmentation network on pseudo-labels to bootstrap the system on images with multiple objects. We show that while being conceptually simple our proposed baseline is surprisingly strong. We present results on PASCAL VOC that go far beyond the current state of the art (50.0 mIoU), and we report for the first time results on MS COCO for the whole set of 81 classes: our method discovers 34 categories with more than 20% IoU, while obtaining an average IoU of 19.6 for all 81 categories. © 2023 11th International Conference on Learning Representations, ICLR 2023. All rights reserved.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Synthetically Generating Human-like Data for Sequential Decision Making Tasks via Reward-Shaped Imitation Learning

Synthetically Generating Human-like Data for Sequential Deci...

引用

Synthetic Data for Artificial Intelligence and Machine Learning: Tools, Techniques, and Applications 2023

作者： Brandt, Bryan Dasgupta, Prithviraj Distributed Intelligent Systems Section Information Technology Division Naval Research Laboratory WashingtonDC United States Computer Science Department University of Wisconsin-Whitewater WI United States

ISBN: (纸本)9781510661721

We consider the problem of synthetically generating data that can closely resemble human decisions made in the context of an interactive human-AI system like a computer game. We propose a novel algorithm that can generate synthetic, human-like, decision making data while starting from a very small set of decision making data collected from humans. Our proposed algorithm integrates the concept of reward shaping with an imitation learning algorithm to generate the synthetic data. We have validated our synthetic data generation technique by using the synthetically generated data as a surrogate for human interaction data to solve three sequential decision making tasks of increasing complexity within a small computer game-like setup. Different empirical and statistical analyses of our results show that the synthetically generated data can substitute the human data and perform the game-playing tasks almost indistinguishably, with very low divergence, from a human performing the same tasks. © 2023 SPIE.

关键词： Decision making

来源：评论

学校读者我要写书评

暂无评论

Robust-MSA: Understanding the Impact of Modality Noise on Multimodal Sentiment Analysis 37

Robust-MSA: Understanding the Impact of Modality Noise on Mu...

引用

37th AAAI Conference on Artificial Intelligence, AAAI 2023

作者： Mao, Huisheng Zhang, Baozheng Xu, Hua Yuan, Ziqi Liu, Yihe State Key Laboratory of Intelligent Technology and Systems Department of Computer Science and Technology Tsinghua University Beijing100084 China Beijing100084 China School of Information Science and Engineering Hebei University of Science and Technology Shijiazhuang050018 China

ISBN: (纸本)9781577358800

Improving model robustness against potential modality noise, as an essential step for adapting multimodal models to real-world applications, has received increasing attention among researchers. For Multimodal Sentiment Analysis (MSA), there is also a debate on whether multimodal models are more effective against noisy features than unimodal ones. Stressing on intuitive illustration and in-depth analysis of these concerns, we present Robust-MSA, an interactive platform that visualizes the impact of modality noise as well as simple defence methods to help researchers know better about how their models perform with imperfect real-world data. Copyright © 2023, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

Autism Spectrum Disorder Detection from EEG Through Hjorth Parameters and Classification Using Neural Network 1st

Autism Spectrum Disorder Detection from EEG Through Hjorth P...

引用

1st International Conference on Machine Intelligence and Emerging Technologies, MIET 2022

作者： Peya, Zahrul Jannat Zaman, Bipasha Akhand, M.A.H. Siddique, Nazmul Department of Computer Science and Engineering Khulna University of Engineering and Technology Khulna9203 Bangladesh School of Computing Engineering and Intelligent Systems Ulster University Northern Ireland Londonderry United Kingdom

ISBN: (纸本)9783031346217

Autism spectrum disorders (ASD) are a collection of neurodevelopmental disorders. Even though ASD has no cure, early detection and intervention can help developing language, behavior, and communication skills. Research shows that ASD can be diagnosed from brain signals, particularly electroencephalography (EEG) where the brain activity is recorded over time as an EEG signal from humans’ scalp and then used to study neuropsychiatric disorders. This study investigates the classification performance of the Neural Networks (NN) model with Hjorth features namely activity, mobility, and complexity extracted from EEG signals through selected channels using the XGBoost algorithm. The classification accuracy is 80% using all 19 channels of EEG data whereas the accuracy of the model using activity, mobility and complexity reached 100% with selected channels. Hjorth parameter-based NN model with channel selection seems to be promising that improves the computation complexity significantly. © 2023, ICST Institute for computer sciences, Social Informatics and Telecommunications Engineering.

关键词： Electroencephalography

来源：评论

学校读者我要写书评

暂无评论

Game Theoretic Decision-Making for Autonomous Driving Considering Traffic Rules at Roundabouts 8

Game Theoretic Decision-Making for Autonomous Driving Consid...

引用

8th CAA International Conference on Vehicular Control and Intelligence, CVCI 2024

作者： Huang, Liting Chu, Duanfeng Deng, Zejian Wang, Rukang Chi, Ao Lu, Liping Intelligent Transportation Systems Research Center Wuhan University of Technology Wuhan China University of Waterloo Department of Mechanical and Mechatronics Engineering Waterloo Canada School of Computer Science and Artificial Intelligence Wuhan University of Technology Wuhan China

ISBN: (纸本)9798331504892

Vehicle interactions at roundabouts are complex, with significant issues arising from right-of-way assignments and the uncertainty of driver behavior. A decentralized game-theoretic decision-making model is proposed, integrating traffic rules to effectively manage complex vehicle interactions in the dynamic traffic scenarios. Linear Temporal Logic (LTL) is used to digitize traffic rules, and vehicle trajectories are extracted from the INTERACTION dataset. These data serve as input for the random forest model to determine the ambiguous thresholds in the digitized traffic rules. These rules are then transformed into constraints and cost functions in the game-theoretic model to enhance decision-making safety and rule compliance. The performances of the proposed decision-making model are validated on the roundabout scenes extracted from the INTERACTION dataset. The results indicate that our model surpasses others in terms of safety, efficiency, and comfort, enabling safe and efficient decision-making in complex traffic scenarios, thus improving the decision-making performance and adaptability of autonomous driving systems at roundabout. © 2024 IEEE.

关键词： Temporal logic

来源：评论

学校读者我要写书评

暂无评论

Multilabel Machine Learning System for Predicting Gender and Age Group from Human Gait 2

Multilabel Machine Learning System for Predicting Gender and...

引用

2nd IEEE International Conference on Interdisciplinary Approaches in Technology and Management for Social Innovation, IATMSI 2024

作者： Qomariyah, Nunung Nurul Nurul Fajar, Ahmad Yono, Vincent Kazakov, Dimitar Bina Nusantara University School of Computing and Creative Arts Computer Science Department Jakarta11480 Indonesia Bina Nusantara University Binus Graduate Program Information System Management Information Systems Management Department Jakarta11480 Indonesia University of York Computer Science Department YO10 5GH United Kingdom

ISBN: (纸本)9798350360523

This paper explores the development of a multilabel machine learning system for predicting both gender and age from human gait patterns. Gait analysis, a non-intrusive method of identifying subtle nuances in human movement, has proven to be a rich source of information related to demographic characteristics. The research extends beyond traditional single-label classification approaches, adopting a multilabel framework to simultaneously predict gender and age *** study evaluates various multilabel machine learning algorithms, with the Random k-labeLsets (RAKEL) algorithm demonstrating superior performance in predicting gender and age labels from human gait datasets. The accuracy of the algorithm can reach up to 87%. We also compared the multilabel approach with several multiclass algorithms such as Decision Tree, Random Forest, Gradient Boosting, K-Neighbors and XGBoost. However, when we also considering the training time, Classifier Chain algorithm showed the best trade off with the accuracy of 86% and the training time is twice faster than the RAKEL algorithm. © 2024 IEEE.

关键词： Gait analysis

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：