检索结果-内蒙古大学图书馆

arXiv 2025年

作者： Dereich, Steffen Do, Thang Jentzen, Arnulf Weber, Frederic Institute for Mathematical Stochastics Faculty of Mathematics and Computer Science University of Münster Germany School of Data Science The Chinese University of Hong Kong Shenzhen [CUHK-Shenzhen China Deparment of Probability and Statistic Institute of Mathematics Vietnam Academy of Science and Technology Viet Nam School of Data Science Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen [CUHK-Shenzhen China Applied Mathematics: Institute for Analysis and Numerics Faculty of Mathematics and Computer Science University of Münster Germany

Deep learning algorithms – typically consisting of a class of deep artificial neural networks (ANNs) trained by a stochastic gradient descent (SGD) optimization method – are nowadays an integral part in many areas of science, industry, and also our day to day life. Roughly speaking, in their most basic form, ANNs can be regarded as functions that consist of a series of compositions of affine-linear functions with multidimensional versions of so-called activation functions. One of the most popular of such activation functions is the rectified linear unit (ReLU) function R ∋ x 7→ max{x, 0} ∈ R. The ReLU function is, however, not differentiable and, typically, this lack of regularity transfers to the cost function of the supervised learning problem under consideration. Regardless of this lack of differentiability issue, deep learning practioners apply SGD methods based on suitably generalized gradients in standard deep learning libraries like TensorFlow or Pytorch. In this work we reveal an accurate and concise mathematical description of such generalized gradients in the training of deep fully-connected feedforward ANNs and we also study the resulting generalized gradient function analytically. Specifically, we provide an appropriate approximation procedure that uniquely describes the generalized gradient function, we prove that the generalized gradients are limiting Fréchet subgradients of the cost functional, and we conclude that the generalized gradients must coincide with the standard gradient of the cost functional on every open sets on which the cost functional is continuously *** Codes 68T07 Copyright © 2025, The Authors. All rights reserved.

关键词： Cost functions

来源：评论

学校读者我要写书评

暂无评论

Design of Intelligent Environmental Protection Agriculture System

Design of Intelligent Environmental Protection Agriculture S...

引用

Electronic Communications, Internet of Things and big data (ICEIB), International Conference on

作者： Zhensong Ni Shuri Cai Cairong Ni Fujian Polytechnic Normal University School of Big Data and Artificial Intelligence Fujian Province China Ministry of Transport Institute of Highway Science Beijing China

ISBN: (数字)9798350360721

ISBN: (纸本)9798350360738

An intelligent agricultural system was developed to turn on or off the designated equipment based on air temperature, soil temperature, and humidity in a greenhouse and environmental parameters such as light, leaf humidity, and dew point temperature. According to the needs of users, it operates at any time and monitors the ecological information of the facility and environmental parameters through intelligent management. The temperature and humidity of the greenhouse are controlled remotely using sensors of the wireless signal transceiver module. The system also includes an intelligent grain depot system and controls temperature and humidity with the connection of a computer or a mobile phone.

关键词： Temperature sensors Temperature measurement Wireless communication Green products Humidity Solar energy Temperature control

来源：评论

学校读者我要写书评

暂无评论

Multi-condition guided depth map completion method based on diffusion model 4

Multi-condition guided depth map completion method based on ...

引用

4th International Conference on Computer science and Communication technology, ICCSCT 2023

作者： Qian, Ren Qiu, Wenfeng Yang, Wenbang Li, Jianhua Wang, Xin’an Zhao, Yong College of Computer Science and Technology Guizhou University Guizhou Guiyang550025 China College of Artificial Intelligence and Big data Luzhou Vocational and Technical College Sichuan Luzhou646000 China School of Electronic and Computer Engineering Shenzhen Graduate School of Peking University Guangdong Shenzhen518055 China Co. Ltd Guangdong Shenzhen518107 China

ISBN: (纸本)9781510671232

The diffusion model has been widely applied in various aspects of artificial intelligence due to its flexible and diverse generative performance. However, there is a lack of research on applying diffusion models in the field of depth map restoration. This is primarily because the diverse generation capabilities are not sufficient to meet the requirements for depth map completion. Depth map completion requires rational completion in the current scene. In order to adapt diffusion models for depth map completion tasks, this paper proposes the Multi Condition Diffusion Model (MCDM). It allows the addition of conditional information to constrain the model’s rational completion of ill-regions. The MultiConditionLN module effectively adds multiple conditions to the depth map completion task. This module uses the completion region mask and the input image as conditions to constrain the model’s generation process. This enables the model to complete the regions that need restoration based on the scene of the input image. The proposed model achieves promising results on depth map datasets. © 2023 SPIE.

关键词： Restoration

来源：评论

学校读者我要写书评

暂无评论

Gradient Tracking with Multiple Local SGD for Decentralized Non-Convex Learning

Gradient Tracking with Multiple Local SGD for Decentralized ...

引用

IEEE Conference on Decision and Control

作者： Songyang Ge Tsung-Hui Chang School of Science and Engineering The Chinese University of Hong Kong Shenzhen and Shenzhen Research Institute of Big Data Shenzhen China

The stochastic Gradient Tracking (GT) method for distributed optimization, is known to be robust against the inter-client variance caused by data heterogeneity. However, the stochastic GT method can be communication-intensive, requiring a large number of communication rounds of message exchange for convergence. To address this challenge, this paper proposes a new communication-efficient stochastic GT algorithm called the Local Stochastic GT(LSGT) algorithm, which adopts the local stochastic gradient descent (local SGD) technique in the GT method. With LSGT, each agent can perform multiple SGD updates locally within each communication round. Although it is not known previously whether the stochastic GT method can benefit from the local SGD, we establish the conditions under which our proposed LSGT algorithm enjoys the linear speedup brought by local SGD. Compared with the existing work, our analysis requires less restrictive conditions on the mixing matrix and algorithm stepsize. Moreover, it reveals that the local SGD does not only reserve the resilience of the stochastic GT method against the data heterogeneity but also speeds up reducing the tracking error reduction in the optimization process. The experimental results demonstrate that the proposed LSGT exhibits improved convergence speed and robust performance in various heterogeneous environments.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Multi-modal Salient Object Detection via a Unified Diffusion Model

Multi-modal Salient Object Detection via a Unified Diffusion...

引用

2025 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2025

作者： Zhang, Shuo Huang, Jiaming Tang, Wenbing Tian, Lili Wei, Yuang Liu, Jing Shanghai Key Laboratory of Trustworthy Computing East China Normal University Shanghai China Huolala Shenzhen China College of Computing and Data Science Nanyang Technological University Singapore School of Computer Science and Technology East China Normal University China

ISBN: (纸本)9798350368741

Salient Object Detection (SOD) aims to identify and segment the most striking elements within an image. Salient object detection methods can be differentiated into several types according to the input data, such as RGB-D (Depth) and RGB-T (Thermal). Previous research primarily focused on saliency detection for single data types. However, forcing an RGB-D SOD model to process RGB-T data will degrade its performance significantly. In addition, current methods still face challenges in detecting fine edge details of salient objects and achieving end-to-end training. To address these issues, we introduce diffSOD, which leverages stable diffusion and cross-modal feature rectification and fusion module for saliency detection by transforming salient object detection into a denoising process from a noisy mask to an object mask. It offers a unified solution for salient object detection that seamlessly spans both RGB-D SOD and RGB-T SOD. Extensive experiments validate the effectiveness of the proposed diffSOD, demonstrating its ability to efficiently detect salient objects across both RGB-D and RGB-T data, while achieving superior performance over state-of-the-art methods. © 2025 IEEE.

关键词： attention interaction fusion latent diffusion model multi-modal salient object detection

来源：评论

学校读者我要写书评

暂无评论

Research on the Physical Mechanism of Glass-Forming Ability Based on Ensemble Deep Learning and Shap Method

SSRN

引用

SSRN 2023年

作者： Lin, Yuancheng Liang, Yongchao Institute of Advanced Optoelectronic Materials and Technology School of Big Data and Information Engineering Guizhou University Guiyang550025 China

Glass forming ability (GFA) of Metallic Glasses (MGs) is the result of the interaction of multiple factors. Most previous studies have only focused on characteristic temperatures, and established machine learning models have significant limitations due to their inability to fully understand the dynamics and topological structure of MGs. Therefore, three deep learning models are build using thermodynamic parameters such as glass transition temperature Tg, liquidus temperature Tl and onset crystallization temperature Tx, as well as kinetic and topological parameters such as total electronegative TEN, average atomic volume VA, atomic size difference d, and mixing entropy Sm. To forecast GFA and examine the mechanism of physical impacts of different characteristics on GFA, artificial neural network (ANN), convolutional neural network (CNN), and long short-term memory network (LSTM) are utilized. Experimental results show that low Tl and high Tx effectively improve the GFA. Furthermore, high Sm promotes amorphous formation by ANN, and GFA is better when VA is greater than 15 by CNN, and GFA can be improved when the TEN is less than 1.4 or greater than 2.0 by LSTM. Finally, the ensemble deep learning approach is used to aggregate the features learned by the three models, increasing the model's accuracy to 0.8629 and summarizing the critical conditions to be satisfied for the manufacture of bulk amorphous Zr-Cu-Al-Ag alloy. This research not only improves the model's accuracy in predicting GFA, but also investigates the influence of thermodynamic, kinetic, and topological characteristics on GFA, and analyzes the critical conditions corresponding to various bulk MGs. © 2023, The Authors. All rights reserved.

关键词： Glass

来源：评论

学校读者我要写书评

暂无评论

RefXVC: Cross-Lingual Voice Conversion with Enhanced Reference Leveraging

arXiv

引用

arXiv 2024年

作者： Zhang, Mingyang Zhou, Yi Ren, Yi Zhang, Chen Yin, Xiang Li, Haizhou The Shenzhen Research Institute of Big Data School of Data Science The Chinese University of Hong Kong Shenzhen China The Department of Electrical and Computer Engineering National University of Singapore Singapore Speech & Audio Team ByteDance AI Lab

This paper proposes RefXVC, a method for cross-lingual voice conversion (XVC) that leverages reference information to improve conversion performance. Previous XVC works generally take an average speaker embedding to condition the speaker identity, which does not account for the changing timbre of speech that occurs with different pronunciations. To address this, our method uses both global and local speaker embeddings to capture the timbre changes during speech conversion. Additionally, we observed a connection between timbre and pronunciation in different languages and utilized this by incorporating a timbre encoder and a pronunciation matching network into our model. Furthermore, we found that the variation in tones is not adequately reflected in a sentence, and therefore, we used multiple references to better capture the range of a speaker’s voice. The proposed method outperformed existing systems in terms of both speech quality and speaker similarity, highlighting the effectiveness of leveraging reference information in cross-lingual voice conversion. The converted speech samples can be found on the website: http://*** © 2024, CC BY-NC-SA.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

MicGuard: a comprehensive detection system against out-of-band injection attacks for different level microphone-based devices 24

MicGuard: a comprehensive detection system against out-of-ba...

引用

Proceedings of the 33rd USENIX Conference on Security Symposium

作者： Tiantian Liu Feng Lin Zhongjie Ba Li Lu Zhan Qin Kui Ren State Key Laboratory of Blockchain and Data Security Zhejiang University School of Cyber Science and Technology Zhejiang University Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security

ISBN: (纸本)9781939133441

The integration of microphones into sensors and systems, serving as input interfaces to intelligent applications and industrial manufacture, has raised public concerns regarding their input perception. Studies have uncovered the potential dangers posed by out-of-band injection attacks on microphones, encompassing ultrasound, laser, and electromagnetic attacks, injecting commands or interferences for malicious purposes. Despite existing efforts on defense against ultrasound injections, there is a critical gap in addressing the risks posed by other out-of-band injections. To bridge this gap, this paper proposes MicGuard, a comprehensive passive detection system against out-of-band attacks. Without relying on prior information from attacking and victim devices, MicGuard leverages carrier traces and spectral chaos observed by injection phenomena across different levels of devices. The carrier traces are used in a prejudgment to fast reject partial injected signals, and the following memory-based detection model to distinguish anomaly based on the quantified chaotic entropy extracted from publicly available audio datasets. MicGuard is evaluated on a wide range of microphone-based devices including sensors, recorders, smartphones, and tablets, achieving an average AUC of 98% with high robustness and universality.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Image description software generation based on data mining and computer vision

引用

Measurement: Sensors 2024年 33卷

作者： Wang, Tao Zhu, Xiaolin School of Software Changsha Social Work College Hunan Changsha410004 China Alibaba Cloud Big Data Application College Zhuhai College of Science and Technology Guangdong Zhuhai519041 China

In order to solve the problems of low quality of image description, insufficient use of image features, and single level of recurrent neural network in image description generation, this paper proposes an image description generation method based on multi-scale features and computer vision. The algorithm uses the pre trained target detection network to extract the image features on different layers of the convolution neural network, inputs the image features into the multi attention structure layer by layer, connects the multi attention structure with the multi-layer recurrent neural network in turn, and constructs a multi-level image description generation network model. Adding residual connections to multilayer recurrent neural networks can improve the network performance, and can effectively avoid the network degradation caused by network deepening. The experimental results show that in the mscoco test set, the bleu-1 and cider scores of the proposed algorithm can reach 0.804 and 1.167 respectively, which is significantly better than the top-down image description generation algorithm based on a single attention structure. Conclusion: the image description generated by the proposed algorithm can show better image details. © 2024

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

AI-Powered Autonomous Token Analysis and Liquidity Provisioning System

AI-Powered Autonomous Token Analysis and Liquidity Provision...

引用

Pervasive Computing and Social Networking (ICPCSN), International Conference on

作者： Ragul B S Kevin Paul Jacob S. Srividhya B.Tech CSE w/s Blockchain Technology Department of Data Science and Business Systems School of Computing S.R.M Institute of Science and Technology Chennai India Department of Data Science and Business Systems School of Computing S.R.M Institute of Science and Technology Chennai India

ISBN: (数字)9798331535193

ISBN: (纸本)9798331535209

This paper presents an AI-powered autonomous system for token selection and liquidity provisioning within DLMM (Dynamic Liquidity Management Market Maker) pools on the Solana blockchain. The core objective is to optimize real-time liquidity deployment while minimizing risk in volatile DeFi environments. By integrating natural language processing (NLP) for sentiment analysis, time-series forecasting for price prediction, and anomaly detection for abnormal trading patterns, the system offers dynamic, data-driven liquidity decisions. It analyzes social media sentiment (from Twitter, Telegram), on-chain market data (via BirdEye and Meteora APIs), and liquidity trends to identify promising token pairs. The AI agent finds trending tokens and evaluates possible liquidity pools using a dynamic risk-adjusted fee-to-TVL ratio by means of 24-hour volume trends, buy/sell ratios, rug detection metrics, whale/sniper activity, and a computed volume-to-market cap *** a token is chosen, the agent automatically constructs or verifies a token-SOL pair, applies liquidity with an optimal allocation approach spanning $100 to $1,000, and dynamically rebalances positions anomaly detection. When pre-defined market thresholds are activated, the system actively changes liquidity, claims earned fees, and leaves positions to guarantee adaptive risk management in tumultuous times. Experimental findings show in real-world distributed markets the scalability, durability, and efficiency of our AI-enhanced liquidity provisioning system.

关键词： Sentiment analysis Social networking (online) Autonomous systems Market research Real-time systems Blockchains Risk management Forecasting Anomaly detection Testing

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：