检索结果-内蒙古大学图书馆

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Geisler, Simon Kosmala, Arthur Herbst, Daniel Günnemann, Stephan Department of Computer Science Munich Data Science Institute Technical University of Munich Germany

Spatial Message Passing Graph Neural networks (MPGNNs) are widely used for learning on graph-structured data. However, key limitations of -step MPGNNs are that their "receptive field" is typically limited to the -hop neighborhood of a node and that information exchange between distant nodes is limited by oversquashing. Motivated by these limitations, we propose Spatio-Spectral Graph Neural networks (S2GNNs) - a new modeling paradigm for Graph Neural networks (GNNs) that synergistically combines spatially and spectrally parametrized graph filters. Parameterizing filters partially in the frequency domain enables global yet efficient information propagation. We show that S2GNNs vanquish over-squashing and yield strictly tighter approximation-theoretic error bounds than MPGNNs. Further, rethinking graph convolutions at a fundamental level unlocks new design spaces. For example, S2GNNs allow for free positional encodings that make them strictly more expressive than the 1-Weisfeiler-Leman (WL) test. Moreover, to obtain general-purpose S2GNNs, we propose spectrally parametrized filters for directed graphs. S2GNNs outperform spatial MPGNNs, graph transformers, and graph rewirings, e.g., on the peptide long-range benchmark tasks, and are competitive with state-of-the-art sequence modeling. On a 40 GB GPU, S2GNNs scale to millions of nodes. © 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

REVISITING ROBUSTNESS IN GRAPH MACHINE LEARNING 11

REVISITING ROBUSTNESS IN GRAPH MACHINE LEARNING

引用

11th International Conference on Learning Representations, ICLR 2023

作者： Gosch, Lukas Sturm, Daniel Geisler, Simon Günnemann, Stephan Department of Computer Science Munich Data Science Institute Technical University of Munich Germany

Many works show that node-level predictions of Graph Neural networks (GNNs) are unrobust to small, often termed adversarial, changes to the graph structure. However, because manual inspection of a graph is difficult, it is unclear if the studied perturbations always preserve a core assumption of adversarial examples: that of unchanged semantic content. To address this problem, we introduce a more principled notion of an adversarial graph, which is aware of semantic content change. Using Contextual Stochastic Block Models (CSBMs) and real-world graphs, our results uncover: i) for a majority of nodes the prevalent perturbation models include a large fraction of perturbed graphs violating the unchanged semantics assumption;ii) surprisingly, all assessed GNNs show over-robustness - that is robustness beyond the point of semantic change. We find this to be a complementary phenomenon to adversarial examples and show that including the label-structure of the training graph into the inference process of GNNs significantly reduces over-robustness, while having a positive effect on test accuracy and adversarial robustness. Theoretically, leveraging our new semantics-aware notion of robustness, we prove that there is no robustness-accuracy tradeoff for inductively classifying a newly added node. © 2023 11th International Conference on Learning Representations, ICLR 2023. 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

SAMPLING-FREE INFERENCE FOR AB-INITIO POTENTIAL ENERGY SURFACE networks 11

SAMPLING-FREE INFERENCE FOR AB-INITIO POTENTIAL ENERGY SURFA...

引用

11th International Conference on Learning Representations, ICLR 2023

作者： Gao, Nicholas Günnemann, Stephan Department of Computer Science Munich Data Science Institute Technical University of Munich Germany

Recently, it has been shown that neural networks not only approximate the ground-state wave functions of a single molecular system well but can also generalize to multiple geometries. While such generalization significantly speeds up training, each energy evaluation still requires Monte Carlo integration which limits the evaluation to a few geometries. In this work, we address the inference shortcomings by proposing the Potential learning from ab-initio networks (PlaNet) framework, in which we simultaneously train a surrogate model in addition to the neural wave function. At inference time, the surrogate avoids expensive Monte-Carlo integration by directly estimating the energy, accelerating the process from hours to milliseconds. In this way, we can accurately model high-resolution multi-dimensional energy surfaces for larger systems that previously were unobtainable via neural wave functions. Finally, we explore an additional inductive bias by introducing physically-motivated restricted neural wave function models. We implement such a function with several additional improvements in the new PESNet++ model. In our experimental evaluation, PlaNet accelerates inference by 7 orders of magnitude for larger molecules like ethanol while preserving accuracy. Compared to previous energy surface networks, PESNet++ reduces energy errors by up to 74 %. © 2023 11th International Conference on Learning Representations, ICLR 2023. All rights reserved.

关键词： Wave functions

来源：评论

学校读者我要写书评

暂无评论

Artificial intelligence algorithms for object detection and recognition in video and images

引用

Multimedia Tools and Applications 2025年 1-18页

作者： Dakshinamoorthy, Prabakar Rajaram, Gnanajeyaraman garg, Shruti Murugan, Prabhu Manimaran, A. Sundar, Ramesh Department of Data Science and Business System School of Computing SRM Institute of Science and Technology SRM Nagar Kattankulathur Chennai India Department of Computer Science Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Tamilnadu Chennai India Birla Institute of Technology Mesra Ranchi India Department of ECE Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Chennai602105 India Department of Computer Science Engineering Saveetha School Of EngineeringSaveetha Institute of Medical and Technical Sciences Chennai6021055 India Department of Netwoking and Communication School of Computing SRM Institute of Science and Technology SRM Nagar Kattankulathur Chennai India

The usage of machine learning and deep learning algorithms have necessitated Artificial Intelligence'. AI is aimed at automating things by limiting human interference. It is widely used in IT, healthcare, finance, and agriculture. It is achieved through several deep learning algorithms that reflect the human brain's intelligence. These AI algorithms can be manipulated according to changing needs and improved efficiency. This paper tries to utilize the developments made in AI technology to classify the images and recognize the objects present in them. One widely used AI algorithm is CNN (Convolutional Neural networks). The CNN is a deep learning-based algorithm that consists of various layers that extract and filters the parameters present in the images. Some additional layers of ResNet50 and the CNN algorithm are used to extract the parameters to improve image recognition accuracy. The image dataset taken for training and testing the proposed model is imageNet. The images are initially processed before sending them to the proposed model. The proposed model is trained, validated, and tested through the images obtained after the initial processing. The same process is repeated several times until getting the maximum accuracy. The accuracy of the proposed model in terms of image recognition is recorded. The obtained results are compared with other image classification algorithms like VGG16 and VGG19. It is concluded that the proposed model outperforms other traditional methods in terms of accuracy. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2025.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Unified Mechanism-Specific Amplification by Subsampling and Group Privacy Amplification 38

Unified Mechanism-Specific Amplification by Subsampling and ...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Schuchardt, Jan Stoian, Mihail Kosmala, Arthur Günnemann, Stephan Dept. of Computer Science Munich Data Science Institute Technical University of Munich Germany Dept. of Engineering University of Technology Nuremberg Germany

Amplification by subsampling is one of the main primitives in machine learning with differential privacy (DP): Training a model on random batches instead of complete datasets results in stronger privacy. This is traditionally formalized via mechanism-agnostic subsampling guarantees that express the privacy parameters of a subsampled mechanism as a function of the original mechanism's privacy parameters. We propose the first general framework for deriving mechanism-specific guarantees, which leverage additional information beyond these parameters to more tightly characterize the subsampled mechanism's privacy. Such guarantees are of particular importance for privacy accounting, i.e., tracking privacy over multiple iterations. Overall, our framework based on conditional optimal transport lets us derive existing and novel guarantees for approximate DP, accounting with Rényi DP, and accounting with dominating pairs in a unified, principled manner. As an application, we analyze how subsampling affects the privacy of groups of multiple users. Our tight mechanism-specific bounds outperform tight mechanism-agnostic bounds and classic group privacy results. © 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

Adversarial Training for Graph Neural networks: Pitfalls, Solutions, and New Directions 37

Adversarial Training for Graph Neural Networks: Pitfalls, So...

引用

37th Conference on Neural Information Processing Systems, NeurIPS 2023

作者： Gosch, Lukas Geisler, Simon Sturm, Daniel Charpentier, Bertrand Zügner, Daniel Günnemann, Stephan Department of Computer Science Munich Data Science Institute Technical University of Munich Germany Microsoft Research United States

ISBN: (纸本)9781713899921

Despite its success in the image domain, adversarial training did not (yet) stand out as an effective defense for Graph Neural networks (GNNs) against graph structure perturbations. In the pursuit of fixing adversarial training (1) we show and overcome fundamental theoretical as well as practical limitations of the adopted graph learning setting in prior work;(2) we reveal that flexible GNNs based on learnable graph diffusion are able to adjust to adversarial perturbations, while the learned message passing scheme is naturally interpretable;(3) we introduce the first attack for structure perturbations that, while targeting multiple nodes at once, is capable of handling global (graph-level) as well as local (node-level) constraints. Including these contributions, we demonstrate that adversarial training is a state-of-the-art defense against adversarial structure perturbations. © 2023 Neural information processing systems foundation. All rights reserved.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Investigation on Nail Denting or Crumbling from Nail Psoriasis Patient Ailment data using the Machine Learning Algorithm and data Mining Techniques 8

Investigation on Nail Denting or Crumbling from Nail Psorias...

引用

8th International Conference on Electronics, Communication and Aerospace Technology, ICECA 2024

作者： Janani, K. Santhoshkumar, S.P. Vikram, D. Priya, S. Vijayaraj, M. Priyanga, M. Manipal Institute of Technology Department of Data Science and Computer Applications Manipal India Vel Tech Rangarajan Dr. Sagunthala R&d Institute of Science and Technology Avadi Department of Computer Science and Engineering Chennai India Rathinam Technical Campus Department of Artificial Intelligence and Data Science Coimbatore India Rathinam Technical Campus Department of Information Technology Coimbatore India

ISBN: (纸本)9798350367904

Nail denting or crumbling may result from nail psoriasis. A nail condition known as psoriatic onychodystrophy or psoriatic nails. Psoriasis sufferers frequently experience it;reported occurrences range from 10% to 78%. Psoriatic nails are more common in the elderly and in people with psoriatic arthritis. By examining the nails on a human hand, numerous illnesses can be found early in the diagnosing process. The nail color of an individual can help in the diagnosis of specific medical disorders. In this case, the recommended method results in the decision-making process for sickness diagnosis. The system is fed by human nail art. The technique recognizes nail features unique to a given condition by analyzing images of nails. The suggested technique makes advantage of the various distinctive characteristics of human nails to detect illness by changing the colour of the nails. The first training set data, which is derived from an image of a patient's nails with a particular ailment, is handled by the Weka tool. To get the desired results, the feature discoveries of the picture are compared to the training dataset. The distortion of the nail unit is known as nail disease. Nail units have their own disease class due to their own indications, symptoms, causes, and consequences, which may or may not be related to other medical disorders. Nail problems are still difficult and ambiguous to identify. A novel machine learning method for recognising and classifying nail problems from photographs is presented in this study. This framework combines CNN models with data mining techniques to extract characteristics. This research was also contrasted with certain other province algorithms (Support vector, ANN, K - nearest neighbors, and RF) evaluated on datasets and showed positive results. The accuracy numbers show how well the model can distinguish nail conditions. This ground-breaking discovery lays the groundwork for precise diagnosis in dermatology, medical image processing, an

关键词： Medical image processing

来源：评论

学校读者我要写书评

暂无评论

Performance analysis of various classification algorithms for providing competency training to workplace risk prevention

引用

Multimedia Tools and Applications 2025年第15期84卷 15123-15149页

作者： Garg, Shruti Murugan, Prabhu Manimaran, A. Sundar, Ramesh Dakshinamoorthy, Prabakar Rajaram, Gnanajeyaraman Birla Institute of Technology Ranchi Mesra India Department of ECE Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Saveetha University Chennai602105 India Department of Computer Science and Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Saveetha University Chennai602105 India Department of Networking and Communication School of Computing SRM Institute of Science and Technology SRM Nagar Kattankulathur India Department of Data Science and Business Systems School of Computing SRM Institute of Science and Technology SRM Nagar Kattankulathur India

In the workplace, risk prevention helps detect the risks and prevent accidents. To achieve this, workers' mental and physical parameters related to their health should be focused on and analyzed. It helps improve the workers' mental health and thus improves organizational performance. There are a lot of mental illnesses due to excess work and a lack of professional knowledge. It affects the performance of the worker. So, the organization needs to conduct competency training programs to improve their professional skills and the mindset to tackle the work pressure. There are several competency training programs given in various organizations to mold the workers, and it considerably reduces workplaces risks. Competency training in the workplace teaches integrity, honesty, managerial skills, and other workers. Developing organizational skills is the primary concern in this competency training. This paper enabled one of the best machine learning algorithms to be chosen to predict the hazards in the workplace and provide a healthy environment to the workers. For that, several classification methods such as Logistic Regression, Decision trees, AdaBoost classifier, Random Forest, Gradient Boosting classifier, and XGBoost classifier are implemented experimented with, and the performance is verified with python software. From the comparison, it has been found that the XGBoost classifier is selected as the best classifier and identified that it is highly suitable for analyzing and predicting workplace risks data. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Occupational risks

来源：评论

学校读者我要写书评

暂无评论

A Drug Box Recognition System Based on Deep Learning and Cloud-Edge Collaboration for Pharmacy Automation 29

A Drug Box Recognition System Based on Deep Learning and Clo...

引用

29th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2023

作者： Zhu, Honglei Dong, Anming Yu, Jiguo School of Computer Science and Technology Jinan China Big Data Institute Jinan China Shandong Fundamental Research Center for Computer Science Shandong Provincial Key Laboratory of Computer Networks Jinan China

ISBN: (纸本)9798350330717

Drug box recognition is an integral part of the pharmacy automation system (PAS), which checks for discrepancies between the drugs provided by the system and the doctor's prescription with the help of modern image recognition techniques. Although deep learning has facilitated the development of pharmacy automation systems, practical applications still face challenges. First, pharmacies update drugs frequently, and many pharmacies across the country are trained independently, which requires consuming massive computational resources. Secondly, the equipment performance of pharmacies is poor, and it's hard to guarantee the real-time and accuracy of recognition. Therefore, we propose a cloud-side collaborative drug box recognition system. When new drugs are added, the new model is trained in the cloud and uniformly distributed to each edge through this architecture. In terms of real-time recognition, an improved lightweight YOLOv5 model is added to the edge part, and a text recognition module is deployed in the cloud to validate the recognition results at the edge. This cloud-edge collaboration architecture not only ensures the accuracy of recognition, but also reduces the burden on the compute edge, ensures the real-time *** results show that the proposed cloud-edge architecture reduces the recognition latency by 37.18% and achieves a precision of 97.47%. The improved YOLOv5 model reduces GFLOPs by 45% while achieving a 1.6% precision improvement. © 2023 IEEE.

关键词： Cloud-Edge Collaboration Image Recognition Pharmacy Automation Text Recognition

来源：评论

学校读者我要写书评

暂无评论

A Trajectory Tracking System for Zebrafish Based on Embedded Edge Artificial Intelligence 29

A Trajectory Tracking System for Zebrafish Based on Embedded...

引用

29th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2023

作者： Zang, Chuanhao Dong, Anming Yu, Jiguo School of Computer Science and Technology Jinan China Big Data Institute Jinan China Shandong Fundamental Research Center for Computer Science Shandong Provincial Key Laboratory of Computer Networks Jinan China

ISBN: (纸本)9798350330717

Trajectory tracking of zebrafish is an important requirement in studying neurological disorders and developing new psychoactive drug. However, many challenges emerge for stable tracking, since zebrafish are similar in appearance, occlusion, agile, non-linear in moving and easy to swarm, all of which will lead to mistrack for multiple fish. And there is no precedent for tracking zebrafish through embedded edge artificial intelligence device. To overcome these difficulties, we present a tracking system for zebrafish based on RK3588-S. First, We construct an embedded edge AI hardware system consisting of two cameras driven by the RK3588-S, one for the front view and the other for top view of the fish. Then, we develop a 2D tracking algorithm based on YOLOv5 and the Observation-Centric Simple online and real-time tracking (OC-SORT) algorithm, which are transplanted to the RK3588-S for tracking the top and front views of the fish at the edge device. Compared with the previous methods, our method has fewer ID exchanges and highly real-time. Moreover, we apply the MQTT mechanism to establish communication links between the edge and the cloud to reliably transmit data to the cloud. The correspondence between cloud server and embedded AI is 1: N. Finally, we design a multi-view data fusion association algorithm to fuse the data of the two views in the cloud, which are further utilized to build the 3D tracklets of the zebrafish. © 2023 IEEE.

关键词： A Tracking System for Zebrafish Embedded edge artificial intelligence RK3588-S Zebrafish

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：