检索结果-内蒙古大学图书馆

Proceedings of the 36th International Conference on Neural Information Processing Systems

作者： Shengjie Luo Shanda Li Shuxin Zheng Tie-Yan Liu Liwei Wang Di He National Key Laboratory of General Artificial Intelligence School of Intelligence Science and Technology Peking University and Zhejiang Lab Machine Learning Department School of Computer Science Carnegie Mellon University Microsoft Research National Key Laboratory of General Artificial Intelligence School of Intelligence Science and Technology Peking University and Center for Data Science Peking University National Key Laboratory of General Artificial Intelligence School of Intelligence Science and Technology Peking University

ISBN: (纸本)9781713871088

Relative Positional Encoding (RPE), which encodes the relative distance between any pair of tokens, is one of the most successful modifications to the original Transformer. As far as we know, theoretical understanding of the RPE-based Transformers is largely unexplored. In this work, we mathematically analyze the power of RPE-based Transformers regarding whether the model is capable of approximating any continuous sequence-to-sequence functions. One may naturally assume the answer is in the affirmative—RPE-based Transformers are universal function approximators. However, we present a negative result by showing there exist continuous sequence-to-sequence functions that RPE-based Transformers cannot approximate no matter how deep and wide the neural network is. One key reason lies in that most RPEs are placed in the softmax attention that always generates a right stochastic matrix. This restricts the network from capturing positional information in the RPEs and limits its capacity. To overcome the problem and make the model more powerful, we first present sufficient conditions for RPE-based Transformers to achieve universal function approximation. With the theoretical guidance, we develop a novel attention module, called Universal RPE-based (URPE) Attention, which satisfies the conditions. Therefore, the corresponding URPE-based Transformers become universal function approximators. Extensive experiments covering typical architectures and tasks demonstrate that our model is parameter-efficient and can achieve superior performance to strong baselines in a wide range of applications.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Representation Matters for Mastering Chess: Improved Feature Representation in AlphaZero Outperforms Switching to Transformers

arXiv

引用

arXiv 2023年

作者： Czech, Johannes Blüml, Jannis Kersting, Kristian Steingrimsson, Hedinn Artificial Intelligence and Machine Learning Lab TU Darmstadt Germany Darmstadt Germany Centre for Cognitive Science TU Darmstadt Germany Darmstadt Germany Department of Electrical and Computer Engineering Rice University HoustonTX United States HoustonTX United States

While transformers have gained recognition as a versatile tool for artificial intelligence (AI), an unexplored challenge arises in the context of chess — a classical AI benchmark. Here, incorporating Vision Transformers (ViTs) into AlphaZero is insufficient for chess mastery, mainly due to ViTs’ computational limitations. The attempt to optimize their efficiency by combining MobileNet and NextViT outperformed AlphaZero by about 30 Elo. However, we propose a practical improvement that involves a simple change in the input representation and value loss functions. As a result, we achieve a significant performance boost of up to 180 Elo points beyond what is currently achievable with AlphaZero in chess. In addition to these improvements, our experimental results using the Integrated Gradient technique confirm the effectiveness of the newly introduced features. © 2023, CC BY.

关键词： Distribution transformers

来源：评论

学校读者我要写书评

暂无评论

Securing the Digital Data using SFLA based Deep Convolutional Neural Network in Medical Network Environment

Securing the Digital Data using SFLA based Deep Convolutiona...

引用

Data Science and Information System (ICDSIS), IEEE International Conference on

作者： G N Keshava Murthy N Achyutha Prasad Piyush Kumar Pareek Pankaj Zanke Madhu Patil Department of Electronics and Instrumentation Engineering Siddaganga Institute of Technology Tumakuru Karnataka India Department of Computer Science and Engineering East West Institute of Technology Bengaluru India Artificial Intelligence and Machine Learning Nitte Meenakshi Institute of Technology Bengaluru India Senior Technical Business Analyst Sapiens International Corporation Department of ECE BGS College of Engineering and Technology Bangalore India

ISBN: (数字)9798350381665

ISBN: (纸本)9798350381672

As a result of the rapid development of multimedia in relation to network knowledge, accessing digital material has become relatively straightforward. Therefore, more emphasis on picture watermarking is required for IP protection. For this reason, many picture watermarking techniques were developed; nevertheless, they suffer from a lack of transparency and are not as sturdy. Multimedia copyright protection is very important in the field of digital watermarking. Digital picture watermarking is the process of secretly removing a watermark and inserting it into a carrier image. The success of digital watermarking in protecting digital data has attracted a lot of academic interest recently. There has been a lot of recent focus on using deep learning networks in tandem with wavelet-oriented strategies for picture watermarking. However, these benefits cannot be provided by traditional watermarking methods, which need manual methods for both extraction and embedding. In light of this problem, this research is driven to present a unique deep learning-based wavelet-based method for producing watermarked pictures with increased invisibility. A newly created Convolutional Neural Networks (CNN) model with medical pictures, and the deep features are retrieved utilising its fully connected layers. Next, the Shuffled frog-leaping algorithm (SFLA) is used to train the model, which ultimately results in higher quality classifications. The experimental results validate that the suggested model outperformed the baseline on a number of different measures, PSNR. In F $(228 \times 344)$ dataset thevWatermark1 ratio the speckle noise reach as 0.9950 and then gaussian noise reach as 0.9924 and then salt and peper noise reaches as 0.9972 respectively. And then the Watermark2 ratio the speckle noise reach as 0.9950 and then gaussian noise reach as 0.9924 and then salt and peper noise reaches as 0.9972 respectively.

关键词： Correlation coefficient Wavelet transforms Explainable AI Gaussian noise Watermarking Speckle Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Design of a Smart Cloud Controlled based artificial Ventilator Device Controller using Patients Vital Information in a Cloud Data Logs 6

Design of a Smart Cloud Controlled based Artificial Ventilat...

引用

6th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2022

作者： Senthil Kumar, T. Narendra, M. Arul, U. Kavitha, S. Srinivas, Porandla Vit University School of Information Technology and Engineering Tamil Nadu Vellore India Cmr EngineeringCollege Departmentof InformationTechnology Telangana Hyderabad India Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Saveetha Nagar Thandalam Department of Artificial Intelligence and Machine Learning Tamil Nadu Chennai India Hindusthan Institute of Technology Department of Electronics Nd Communication Engineering Tamilnadu Coimbatore India Malla Reddy Institute of Engineering and Technology Department of Information Technology Telangana Secunderabad India

ISBN: (数字)9781665469418

ISBN: (纸本)9781665469418

To breathe, the human lungs employ the reverse pressure created by the diaphragm's contraction motion to suck in air. A ventilator uses a contrary motion to inflate the lungs via a pumping action. A ventilator mechanism must be capable of delivering between 10 and 30 breaths per minute, with the option to regulate ascending increments in sets of two. In addition, the ventilator must be able to alter the amount of air forced into the lungs with each breath. The final but not least option is to change the time length for the inhale to exhalation ratio. Aside from that, the respirator must be able to concurrently regulate the patient's blood oxygen concentration and exhaled pulmonary pressure to prevent over/under air pressure. The ventilator we design and construct using arduino meets all of these parameters in order to provide a dependable but economical ventilator to aid in pandemic situations. To push the ventilation bag, we employ a silicon ventilation bag linked to a servo motor and a one-side push mechanism. © 2022 IEEE.

关键词： Microcontrollers

来源：评论

学校读者我要写书评

暂无评论

Exploring Deep Neural Networks for Accurate Diabetic Retinopathy Prediction

Exploring Deep Neural Networks for Accurate Diabetic Retinop...

引用

International Conference on Communication, Computing and Internet of Things (IC3IoT)

作者： T. Kujani P. Arivubrakan K. Raju P. Mani Sankar B. Surekha Reddy E. Bhavana Computer Science and Engineering Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Chennai India Artificial Intelligence and Machine Learning Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Chennai India Artificial Intelligence and Data Science Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Chennai India

ISBN: (数字)9798350352689

ISBN: (纸本)9798350352696

Diabetic Retinopathy (DR) emerges as a serious complication associated with diabetes mellitus, causing damage to retinal blood vessels, and resulting in vision loss. The critical need for early detection and effective management to curb the progression towards vision impairment and potential blindness underscores the significance of this issue. To address these challenges, have propose a novel deep learning system utilizing a Densely connected Convolutional Network (DenseNet-201) for the efficient detection of DR. This model is specifically trained to classify fundus images into six distinct categories: no DR, mild, moderate, and severe, proliferative DR, and hard exudates. The datasets utilized in this study include the publicly available Diabetic Retinopathy Detection 2015 and Aptos 2019 blindness detection datasets from Kaggle. The proposed system comprises multiple stages, including Data Collection, Preprocessing, Augmentation, and Modeling. During the preprocessing phase, normalization techniques are to eliminate biases or distortions present in the fundus images. Furthermore, the augmentation of data is employed to enhance the diversity and quantity of the training dataset. In the modeling stage, have designed and implemented the deep learning model, with the robust and high-performing DenseNet-201 architecture selected as the model’s backbone. The model exhibits a significant accuracy of 87.7% in its classification task. Additionally, have introduced a Deep Neural Network (DNN) model, achieving an impressive accuracy rate of 95.8%. These results underscore the effectiveness of our proposed system in accurately and efficiently detecting DR. The primary objective of this research is to develop a reliable and automated system for Diabetic Retinopathy detection, with the overarching goal of improving the quality of life for individuals affected by diabetes.

关键词： Deep learning Training Diabetic retinopathy Transfer learning Visual impairment Blindness Robustness

来源：评论

学校读者我要写书评

暂无评论

Contrastive learning for Adapting Language Model to Sequential Recommendation

Contrastive Learning for Adapting Language Model to Sequenti...

引用

IEEE International Conference on Data Mining (ICDM)

作者： Fei-Yao Liang Wu-Dong Xi Xing-Xing Xing Wei Wan Chang-Dong Wang Min Chen Mohsen Guizani School of Computer Science and Engineering Sun Yat-sen University Guangzhou China Contribution during internship at NetEase Games UX Center NetEase Games Guangzhou China Guangdong Provincial Key Laboratory of Intellectual Property and Big Data Guangzhou China School of Computer Science and Engineering South China University of Technology Guangzhou China Pazhou Laboratory Guangzhou China Machine Learning Department Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) Abu Dhabi UAE

ISBN: (数字)9798331506681

ISBN: (纸本)9798331506698

With the explosive growth of information, recommendation systems have emerged to alleviate the problem of information overload. In order to improve the performance of recommendation systems, many existing methods introduce Large Language Models to extract textual information from description text. However, Large Language Models are trained on large-scale generic textual data and may face a semantic gap for downstream recommendation tasks. To address the above issues, we propose Contrastive learning for Adapting Language Model to Sequential Recommendation (CLA-Rec). In CLA-Rec, we first extract text embeddings from description text using Large Language Models and align the text embeddings learned by Large Language Models with the collaborative information through contrastive learning to obtain high-quality item representations. Through semantic alignment, we bridge the semantic gap between Large Language Models and the recommendation task. To map textual information and collaborative information into user representations, we utilize a Transformer model to learn user representations and capture user preferences by combining the semantically aligned item representations. Extensive experiments on three public datasets demonstrate that our method outperforms state-of-the-art approaches on multiple evaluation metrics, illustrating the effectiveness of the CLA-Rec model in adapting Large Language Models to recommendation tasks.

关键词： Measurement Adaptation models Large language models Semantics Collaboration Contrastive learning Transformers Data models Data mining Recommender systems

来源：评论

学校读者我要写书评

暂无评论

RecCoder: Reformulating Sequential Recommendation as Large Language Model-Based Code Completion

RecCoder: Reformulating Sequential Recommendation as Large L...

引用

IEEE International Conference on Data Mining (ICDM)

作者： Kai-Huang Lai Wu-Dong Xi Xing-Xing Xing Wei Wan Chang-Dong Wang Min Chen Mohsen Guizani School of Computer Science and Engineering Sun Yat-sen University Guangzhou China Contribution during internship at NetEase Games UX Center NetEase Games Guangzhou China Guangdong Provincial Key Laboratory of Intellectual Property and Big Data Guangzhou China School of Computer Science and Engineering South China University of Technology Guangzhou China Pazhou Laboratory Guangzhou China Machine Learning Department Mohamed Bin Zayed University of Artificial Intelligence (MBZUAI) Abu Dhabi UAE

ISBN: (数字)9798331506681

ISBN: (纸本)9798331506698

In the evolving landscape of sequential recommendation systems, the application of Large Language Models (LLMs) is increasingly prominent. However, current attempts typically utilize general-purpose LLMs, which present a mismatch in capability and a large semantic gap relative to the specialized needs of recommendation tasks. To tackle these issues, we introduce RecCoder, an innovative model that reformulates sequential recommendation as a code completion task. This approach leverages the superior reasoning capability of code LLMs as a backbone, aligning well with the requirements of recommendation systems. To bridge the semantic gap, RecCoder creates extra tokens for each item and employs item content to initialize token embeddings. Furthermore, we have developed a suite of Semantic Adaptation Fine-tuning tasks, tailored to enhance the model's acquisition of both content and collaborative semantic information, thus aligning the model's intrinsic capabilities with the unique demands of recommendation tasks. Through extensive testing on three public datasets, RecCoder has shown remarkable improvements over existing models in terms of recommendation accuracy and efficiency. This success highlights the substantial yet previously underexplored potential of code LLMs in improving recommendation accuracy and efficiency, suggesting a promising new direction for future research in this area. The implementation code is accessible at https://***/AllminerLab/Code-for-RecCoder-master.

关键词： Adaptation models Codes Accuracy Large language models Semantics Cognition Data models Data mining Recommender systems Testing

来源：评论

学校读者我要写书评

暂无评论

Is L2 physics-informed loss always suitable for training physics-informed neural network? 22

Is L2 physics-informed loss always suitable for training phy...

引用

Proceedings of the 36th International Conference on Neural Information Processing Systems

作者： Chuwei Wang Shanda Li Di He Liwei Wang School of Mathematical Sciences Peking University Machine Learning Department School of Computer Science Carnegie Mellon University and Zhejiang Lab National Key Laboratory of General Artificial Intelligence School of Intelligence Science and Technology Peking University National Key Laboratory of General Artificial Intelligence School of Intelligence Science and Technology Peking University and Center for Data Science Peking University

ISBN: (纸本)9781713871088

The Physics-Informed Neural Network (PINN) approach is a new and promising way to solve partial differential equations using deep learning. The L2 Physics- Informed Loss is the de-facto standard in training Physics-Informed Neural Networks. In this paper, we challenge this common practice by investigating the relationship between the loss function and the approximation quality of the learned solution. In particular, we leverage the concept of stability in the literature of partial differential equation to study the asymptotic behavior of the learned solution as the loss approaches zero. With this concept, we study an important class of high-dimensional non-linear PDEs in optimal control, the Hamilton-Jacobi- Bellman (HJB) Equation, and prove that for general Lp Physics-Informed Loss, a wide class of HJB equation is stable only if p is sufficiently large. Therefore, the commonly used L2 loss is not suitable for training PINN on those equations, while L∞ loss is a better choice. Based on the theoretical insight, we develop a novel PINN training algorithm to minimize the L∞ loss for HJB equations which is in a similar spirit to adversarial training. The effectiveness of the proposed algorithm is empirically demonstrated through experiments.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Stable Heterogeneous Treatment Effect Estimation across Out-of-Distribution Populations

arXiv

引用

arXiv 2024年

作者： Zhang, Yuling Wu, Anpeng Kuang, Kun Du, Liang Sun, Zixun Wang, Zhi Tsinghua Shenzhen International Graduate School Tsinghua University China Tsinghua-Berkeley Shenzhen Institute Tsinghua University China College of Computer Science and Technology Zhejiang University China Machine Learning Department Mohamed bin Zayed University of Artificial Intelligence United Arab Emirates Interactive Entertainment Group Tencent

Heterogeneous treatment effect (HTE) estimation is vital for understanding the change of treatment effect across individuals or subgroups. Most existing HTE estimation methods focus on addressing selection bias induced by imbalanced distributions of confounders between treated and control units, but ignore distribution shifts across populations. Thereby, their applicability has been limited to the in-distribution (ID) population, which shares a similar distribution with the training dataset. In real-world applications, where population distributions are subject to continuous changes, there is an urgent need for stable HTE estimation across out-of-distribution (OOD) populations, which, however, remains an open problem. As pioneers in resolving this problem, we propose a novel Stable Balanced Representation learning with Hierarchical-Attention Paradigm (SBRL-HAP) framework, which consists of 1) Balancing Regularizer for eliminating selection bias, 2) Independence Regularizer for addressing the distribution shift issue, 3) Hierarchical-Attention Paradigm for coordination between balance and independence. In this way, SBRL-HAP regresses counterfactual outcomes using ID data, while ensuring the resulting HTE estimation can be successfully generalized to out-of-distribution scenarios, thereby enhancing the model's applicability in real-world settings. Extensive experiments conducted on synthetic and real-world datasets demonstrate the effectiveness of our SBRL-HAP in achieving stable HTE estimation across OOD populations, with an average 10% reduction in the error metric PEHE and 11% decrease in the ATE bias, compared to the SOTA methods. Copyright © 2024, The Authors. All rights reserved.

关键词： machine learning

来源：评论

学校读者我要写书评

暂无评论

Bangladeshi Traffic Sign Recognition and Classification using CNN with Different Kinds of Transfer learning through a new (BTSRB) Dataset

Bangladeshi Traffic Sign Recognition and Classification usin...

引用

Distributed Computing and Electrical Circuits and Electronics (ICDCECE), IEEE International Conference on

作者： Md. Abu Sayeed Md. Saiful Islam Md Babul Islam Piyush Kumar Pareek Tanbin Islam Rohan Electrnoics and Telecommunication Engineering Chittagong University of Engineering & Technology Chattogram Bangladesh School of ICT DIMES Lab University of Calabria Rende Italy Artificial Intelligence and Machine Learning Nitte Meenakshi Institute of Technology Bengaluru India Bahari Data Research House Dhaka Bangladesh

An accident is the cry of a lifetime. The role of traffic signs is most important in preventing or mitigating accidents. In many cases, while driving, traffic signs are not visible in the conventional approach; if traffic sign detection and recognition can alert the driver of important sign guidance ahead, it will help in accident mitigation. Traffic sign recognition plays a monumental role in expert systems, such as traffic assistance driving systems and automatic driving systems. The prime purpose of this paper is to design and identify a computer-based system that can spontaneously detect the direction of a road sign. For this research work, we have created our own dataset, which is called the Bangladeshi Traffic Sign Recognition Benchmark (BTSRB) dataset. The dataset, BTSRB, was created by capturing images from different angles and under different parameters and conditions. A total of 7320 images were collected to create this comprehensive database. This dataset called BTSRB all the images collected from Bangladesh. In this paper, we used five different types of models (CNN, Inception V3, MobileNetV2, ResNet50, and VGG16), which are pre-trained on the ImageNet dataset. Later, we finetuned the pre-trained model and used transfer learning. The main challenge of this research is collecting datasets from a country like Bangladesh, where no recognized dataset is available. When compared to another model, the accuracy of this model is greater than 91%. This paper emphasizes the significance of traffic sign recognition in expert systems and the necessity for a well-established dataset in nations where such resources are not readily available.

关键词：

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：