检索结果-内蒙古大学图书馆

12th International Conference on Learning Representations, ICLR 2024

作者： Ye, Feiyang Lyu, Yueming Wang, Xuehao Zhang, Yu Tsang, Ivor W. Department of Computer Science and Engineering Southern University of Science and Technology China Australian Artificial Intelligence Institute University of Technology Sydney Australia Centre for Frontier AI Research Agency for Science Technology and Research Singapore Institute of High Performance Computing Agency for Science Technology and Research Singapore Shanghai Artificial Intelligence Laboratory China School of Computer Science and Engineering Nanyang Technological University Singapore

Multi-objective optimization (MOO) has become an influential framework for various machine learning problems, including reinforcement learning and multitask learning. In this paper, we study the black-box multi-objective optimization problem, where we aim to optimize multiple potentially conflicting objectives with function queries only. To address this challenging problem and find a Pareto optimal solution or the Pareto stationary solution, we propose a novel adaptive stochastic gradient algorithm for black-box MOO, called ASMG. Specifically, we use the stochastic gradient approximation method to obtain the gradient for the distribution parameters of the Gaussian smoothed MOO with function queries only. Subsequently, an adaptive weight is employed to aggregate all stochastic gradients to optimize all objective functions effectively. Theoretically, we explicitly provide the connection between the original MOO problem and the corresponding Gaussian smoothed MOO problem and prove the convergence rate for the proposed ASMG algorithm in both convex and non-convex scenarios. Empirically, the proposed ASMG method achieves competitive performance on multiple numerical benchmark problems. Additionally, the state-of-the-art performance on the black-box multi-task learning problem demonstrates the effectiveness of the proposed ASMG method. © 2024 12th International Conference on Learning Representations, ICLR 2024. All rights reserved.

关键词： Multiobjective optimization

来源：评论

学校读者我要写书评

暂无评论

Multilingual Brain Surgeon: Large Language Models Can be Compressed Leaving No Language Behind 30

Multilingual Brain Surgeon: Large Language Models Can be Com...

引用

Joint 30th International Conference on Computational Linguistics and 14th International Conference on Language Resources and Evaluation, LREC-COLING 2024

作者： Zeng, Hongchuan Xu, Hongshen Chen, Lu Yu, Kai X-LANCE Lab Department of Computer Science and Engineering MoE Key Lab of Artificial Intelligence SJTU AI Institute Shanghai Jiao Tong University Shanghai China Suzhou Laboratory Suzhou China

ISBN: (纸本)9782493814104

Large Language Models (LLMs) have ushered in a new era in Natural Language Processing, but their massive size demands effective compression techniques for practicality. Although numerous model compression techniques have been investigated, they typically rely on a calibration set that overlooks the multilingual context and results in significant accuracy degradation for low-resource languages. This paper introduces Multilingual Brain Surgeon (MBS), a novel calibration data sampling method for multilingual LLMs compression. MBS overcomes the English-centric limitations of existing methods by sampling calibration data from various languages proportionally to the language distribution of the model training datasets. Our experiments, conducted on the BLOOM multilingual LLM, demonstrate that MBS improves the performance of existing English-centric compression methods, especially for low-resource languages. We also uncover the dynamics of language interaction during compression, revealing that the larger the proportion of a language in the training set and the more similar the language is to the calibration language, the better performance the language retains after compression. In conclusion, MBS presents an innovative approach to compressing multilingual LLMs, addressing the performance disparities and improving the language inclusivity of existing compression techniques. The codes are available at: https://***/X-LANCE/MBS. © 2024 ELRA Language Resource Association: CC BY-NC 4.0.

关键词： Calibration

来源：评论

学校读者我要写书评

暂无评论

PerReactor: Offline Personalised Multiple Appropriate Facial Reaction Generation 39

PerReactor: Offline Personalised Multiple Appropriate Facial...

引用

39th Annual AAai Conference on Artificial Intelligence, AAai 2025

作者： Zhu, Hengde Kong, Xiangyu Xie, Weicheng Huang, Xin He, Xilin Liu, Lu Shen, Linlin Zhang, Wei Gunes, Hatice Song, Siyang School of Computing and Mathematical Sciences University of Leicester Leicester United Kingdom Department of Computer Science University of Exeter Exeter United Kingdom Affect AI Anhui China Computer Vision Institute School of Computer Science & Software Engineering Shenzhen University Shenzhen China South China Normal University Foshan China School of Software Technology Zhejiang University Hangzhou China Department of Computer Science and Technology University of Cambridge Cambridge United Kingdom

ISBN: (纸本)157735897X

In dyadic human-human interactions, individuals may express multiple different facial reactions in response to the same/similar behaviours expressed by their conversational partners depending on their personalised behaviour patterns. As a result, frequently-employed reconstruction loss-based strategies lead the training of previous automatic facial reaction generation (FRG) models to not only suffer from the 'one-to-many mapping' problem, but also fail to comprehensively consider the quality of the generated facial reactions. Besides, none of them considered such personalised behaviour patterns in generating facial reactions. In this paper, we propose the first adversarial FRG model training strategy which jointly learns appropriateness and realism discriminators to provide comprehensive task-specific supervision for training the target facial reaction generators, and reformulates the 'one-to-many (facial reactions) mapping' training problem as a 'one-to-one (distribution) mapping' training task, i.e., the FRG model is trained to output a distribution representing multiple appropriate/plausible facial reaction from each input human behaviour. In addition, our approach also serves as the first offline FRG approach that considers personalised behaviour patterns in generating of target individuals' facial reactions. Experiments show that our PerReactor not only largely outperformed all existing offline solutions for generating appropriate, diverse and realistic facial reactions, but also is the first offline approach that can effectively generate personalised appropriate facial reactions. Copyright © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

CLIMS++: Cross Language Image Matching with Automatic Context Discovery for Weakly Supervised Semantic Segmentation

引用

International Journal of computer Vision 2025年 1-20页

作者： Xie, Jinheng Deng, Songhe Hou, Xianxu Luo, Zhaochuan Shen, Linlin Huang, Yawen Zheng, Yefeng Shou, Mike Zheng Computer Vision Institute School of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China Department of Computer Science Wenzhou-Kean University Wenzhou325060 China Guangdong Provincial Key Laboratory of Intelligent Information Processing Shenzhen University Shenzhen China Show Lab National University of Singapore Singapore639798 Singapore School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Suzhou215400 China Jarvis Research Center Tencent YouTu Lab Shenzhen518703 China

While promising results have been achieved in weakly-supervised semantic segmentation (WSSS), limited supervision from image-level tags inevitably induces discriminative reliance and spurious relations between target classes and background regions. Thus, Class Activation Map (CAM) usually tends to activate discriminative object regions and falsely includes lots of class-related backgrounds. Without pixel-level supervisions, it could be very difficult to enlarge the foreground activation and suppress those false activation of background regions. In this paper, we propose a novel framework of Cross Language Image Matching with Automatic Context Discovery (CLIMS++), based on the recently introduced Contrastive Language-Image Pre-training (CLIP) model, for WSSS. The core idea of our framework is to introduce natural language supervision to activate more complete object regions and suppress class-related background regions in CAM. In particular, we design object, background region, and text label matching losses to guide the model to excite more reasonable object regions of each category. In addition, we propose to automatically find spurious relations between foreground categories and backgrounds, through which a background suppression loss is designed to suppress the activation of class-related backgrounds. The above designs enable the proposed CLIMS++ to generate a more complete and compact activation map for the target objects. Extensive experiments on PASCAL VOC 2012 and MS COCO 2014 datasets show that our CLIMS++ significantly outperforms the previous state-of-the-art methods. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Multi Label Classification Based on Ensemble Learning: Investigative Study 25

Multi Label Classification Based on Ensemble Learning: Inves...

引用

25th International Arab Conference on Information Technology, ACIT 2024

作者： Nasir, Ola Alazaidah, Raed Hadi, Wael Samara, Ghassan Al Daoud, Essam Owaida, Suhaila Abu Elrashidi, Ali Elsoud, Esraa Abu Alzoubi, Haneen Zarqa University Faculty of Information Technology Department of Computer Science Zarqa Jordan Zarqa University Faculty of Information Technology Department of Data Science and Ai Zarqa Jordan Petra University Faculty of Information Technology Department of Cyber Security Amman Jordan Zarqa University Faculty of Information Technolog Department of Computer Science Zarqa Jordan Al-al-Bayt University Faculty of Information Technology Department of Computer Science Mafraq Jordan University of Business and Technology Department of Electrical Engineering Jeddah21432 Saudi Arabia Applied Science Private University Faculty of It Department of Computer Science Amman Jordan

ISBN: (纸本)9798331540012

In the big data technology nowadays, activities associated with Multi-Label Classification (mlC) pose big and complex challenges, receiving terrific interest in diverse fields. Existing mlC algorithms suffer from low overall performance. This paper considers four well-known Problem Transformation Methods (PTMs). Two of these PTMs are based on the frequency of the labels (MFL and LFL) while the last two PTMs are based on the positive correlations among labels. These PTMs have been implemented and compared against four multi label datasets of various sizes to identify the best one that fits the nature of mlC datasets. Moreover, six popular ensemble learning models have been carried out and compared on the four datasets with respect to several evaluation metrics such as Accuracy, Hamming loss, and Exact Match and using SMO, DecisionStump, Random Forest, and Random Commission classifiers. The results revealed that MFL accomplished the quality performance amongst the four PTMs, and Stacking SVM is the best ensemble learning technique. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Machine Learning Based Assessment of Elite Football Players Based on Anthropometric and Motor Fitness Parameters with Regard to their Playing Positions

引用

SN computer science 2024年第8期5卷 1-10页

作者： Sneha, Santoshi Prithvi, B.S. Niranjanamurthy, M. Kiran Kumar, H.K. Dayananda, P. MediNXT Technologies Pvt Ltd Bangalore India Department of Computer Science and Engineering Aditya College of Engineering and Technology (affiliated to Visvesvaraya Technological University Belagavi) Bangalore 560089 India Department of AI and ML BMS Institute of Technology and Management (affiliated to Visvesvaraya Technological University Belagavi) Bangalore India Department of Physical Education M S Ramaiah Institute of Technology (affiliated to Visvesvaraya Technological University Belagavi) Bangalore India Department of Information Technology Manipal Institute of Technology Bengaluru Manipal Academy of Higher Education Manipal India

Sports players strive to be the epitome of human excellence, pushing the barrier of skill and execution with training, focus and direction, amplified by regular training and practice. This could be attributed to various factors such as response to stimuli, physical factors, psychological factors etc. The present study incorporated the prediction of the most suited playing position of elite male football players using machine learning approaches based on their Anthropometric Parameters (AP–11 parameters) and Motor Fitness Parameters (MFP–7 parameters). Of the features analysed, results identified the position indicative nature of some parameters among6 AP (Height, Body Mass Index, Basal Metabolic Rate, Fat %, Thigh Circumference, Calf circumference) and 4 MFP (120 m, 80 m, 40 m dash and T-Test) by Spearman’s Rank Correlation Test. Further, the prediction of ideal playing position was achieved using various classifiers such as Support Vector Machine (SVM), SVM with over sampled data, SVM with hyperparameter tuning, SVM with variable scaling and Extreme Gradient Boosting (XG Boost). Among these, the highest classification accuracy and f1-score at 92% and 0.92 respectively were obtained for XG Boost Classifier which portrayed a faster performance as compared to the other approaches. The present study could be useful in professional sports training and rehabilitation so as to help the players perform better in the football game. © The Author(s) 2024.

关键词： Anthropometric parameters Football players Hyperparameter tuning Physiological parameters Spearman’s rank correlation test XG boost classifier

来源：评论

学校读者我要写书评

暂无评论

RDT-1B: A DIFFUSION FOUNDATION MODEL FOR BIMANUAL MANIPULATION

arXiv

引用

arXiv 2024年

作者： Liu, Songming Wu, Lingxuan Li, Bangguo Tan, Hengkai Chen, Huayu Wang, Zhengyi Xu, Ke Su, Hang Zhu, Jun Department of Computer Science & Technology Institute for AI BNRist Center Tsinghua-Bosch Joint ML Center THBI Lab Tsinghua University China

Bimanual manipulation is essential in robotics, yet developing foundation models is extremely challenging due to the inherent complexity of coordinating two robot arms (leading to multi-modal action distributions) and the scarcity of training data. In this paper, we present the Robotics Diffusion Transformer (RDT), a pioneering diffusion foundation model for bimanual manipulation. RDT builds on diffusion models to effectively represent multi-modality, with innovative designs of a scalable Transformer to deal with the heterogeneity of multi-modal inputs and to capture the nonlinearity and high frequency of robotic data. To address data scarcity, we further introduce a Physically Interpretable Unified Action Space, which can unify the action representations of various robots while preserving the physical meanings of original actions, facilitating learning transferrable physical knowledge. With these designs, we managed to pre-train RDT on the largest collection of multi-robot datasets to date and scaled it up to 1.2B parameters, which is the largest diffusion-based foundation model for robotic manipulation. We finally fine-tuned RDT on a self-created multi-task bimanual dataset with over 6K+ episodes to refine its manipulation capabilities. Experiments on real robots demonstrate that RDT significantly outperforms existing methods. It exhibits zero-shot generalization to unseen objects and scenes, understands and follows language instructions, learns new skills with just 1∼5 demonstrations, and effectively handles complex, dexterous tasks. We refer to the project page for the code and videos. Copyright © 2024, The Authors. All rights reserved.

关键词： Robotic arms

来源：评论

学校读者我要写书评

暂无评论

Comprehensive Analysis and Framework for Text and Graphics Separation in Document Images

Comprehensive Analysis and Framework for Text and Graphics S...

引用

Computational Intelligence, Communication Technology and Networking (CICTN), International Conference on

作者： Shweta Singh Sudeep Varshney Ankur Chaudhary Department of CSE Sharda School of Engineering & Technology Sharda University Greater Noida AI/ML Computational Science Manager Accenture Pvt Ltd

ISBN: (数字)9798331530389

ISBN: (纸本)9798331530396

The process of breaking up a digital image into many parts is called segmentation. These sections in scanned papers, are those that have backgrounds, texts, and images. In applications linked to document analysis, text segmentation is a significant issue. The separation of a complex document into non-text and text components is a major challenge in document image analysis. Localization of text in printed document images is a critical processing step for page layout analysis (PLA), specifically for obtaining textual data. Several algorithms have been developed for this subject. However, many algorithms only produce accurate answers for a limited type of documents due to their reliance on specific attributes or assumptions. There is a need to develop effective techniques for many sorts of documents, such as newspapers, magazines, documents, and articles, with arbitrary layouts and nonhomogeneous backgrounds. This study examines the various ways that distinguish between non-text and text in document images.

关键词： Surveys Location awareness Image segmentation Text analysis Image analysis Reviews Digital images Layout Optical character recognition Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Job-SDF: A Multi-Granularity Dataset for Job Skill Demand Forecasting and Benchmarking 38

Job-SDF: A Multi-Granularity Dataset for Job Skill Demand Fo...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Chen, Xi Qin, Chuan Fang, Chuyu Wang, Chao Zhu, Chen Zhuang, Fuzhen Zhu, Hengshu Xiong, Hui University of Science and Technology of China China Computer Network Information Center Chinese Academy of Sciences China Baidu Inc. China Institute of Artificial Intelligence Beihang University China SKLSDE School of Computer Science Beihang University China AI Thrust The Hong Kong University of Science and Technology Guangzhou China Department of Computer Science and Engineering The HongKong University of Science and Technology Hong Kong

In a rapidly evolving job market, skill demand forecasting is crucial as it enables policymakers and businesses to anticipate and adapt to changes, ensuring that workforce skills align with market needs, thereby enhancing productivity and competitiveness. Additionally, by identifying emerging skill requirements, it directs individuals towards relevant training and education opportunities, promoting continuous self-learning and development. However, the absence of comprehensive datasets presents a significant challenge, impeding research and the advancement of this field. To bridge this gap, we present Job-SDF, a dataset designed to train and benchmark job-skill demand forecasting models. Based on millions of public job advertisements collected from online recruitment platforms, this dataset encompasses monthly recruitment demand. Our dataset uniquely enables evaluating skill demand forecasting models at various granularities, including occupation, company, and regional levels. We benchmark a range of models on this dataset, evaluating their performance in standard scenarios, in predictions focused on lower value ranges, and in the presence of structural breaks, providing new insights for further research. Our code and dataset are publicly accessible via the https://***/Job-SDF/benchmark. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Predictive Approach For Assessing Medication Responsiveness in Breast Cancer Through Analysis of Gene Data

A Predictive Approach For Assessing Medication Responsivenes...

引用

Intelligent Control, Computing and Communications (IC3), International Conference on

作者： Kimmi Gupta Jagendra Singh Harender Pratap Singh Manoj Kumar Singh Nitin Gupta Basudeo Singh Roohani School of Computer Science Engineering & Technology Bennett University Greater Noida India Department of CSE (AI & ML) Raj Kumar Goel Institute of Technology Ghaziabad India Computer Science & Engineering IMS Engineering College Ghaziabad India Department of CSE S. D. College of Engineering and Technology Muzaffarnagar India

ISBN: (数字)9798331527495

ISBN: (纸本)9798331527501

Breast cancer remains one of the leading causes of cancer-related deaths worldwide, with treatment responses varying wildly among patients. The inability to predict how an individual patient will respond to a given particular medication is an essential area in optimising treatment plans and better outcomes in such deadly diseases. This research aims to address the challenge of predicting medication responsiveness in breast cancer, analysing genomic data by exploiting publicly available datasets from both The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO). The datasets are comprised of gene expression profiles, somatic mutations, and clinical data such as patient demographics, tumour subtypes, and treatment outcomes. Preprocessing involved cleaning and normalisation of gene expression values, imputation to address missing data, and selection of essential genes that were most relevant for drug response using feature selection techniques. Its dimensionality was reduced using Principal Component Analysis (PCA). Four machine learning models, including artificial neural networks (ANN), Convolutional Neural Networks (CNN), recurrent neural networks with long short-term memory (RNN-LSTM), and autoencoders, are used in this research. All models were trained, validated, and tested against accuracy, precision, recall, F1 score, and AUC-ROC in that the highest performance was achieved at 98.89% accuracy using the CNN model. The findings point towards the fact that deep learning models, especially CNN, open promising avenues for personalised breast cancer treatment. The findings, therefore, hold significant importance for the application of ai in precision medicine, potentially enabling more accurately targeted therapies, minimising side effects, and ultimately improving patient outcomes in oncology.

关键词： Accuracy Recurrent neural networks Precision medicine Genomics Oncology Breast cancer Convolutional neural networks Gene expression Bioinformatics Principal component 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：