检索结果-内蒙古大学图书馆

Tuning-up learning parameters for Deep Convolutional Neural Network:A Case Study for Hand-Drawn Sketch Images

Journal of Electronic Science and Technology 2022年第3期20卷 305-318页

作者： Shaukat Hayat Kun She Muhammad Mateen Parinya Suwansrikham Muhammad Abdullah Ahmed Alghaili the School of Information and Software Engineering University of Electronic Science and Technology of ChinaChengdu 610054 the School of Big Data&Software Engineering Chongqing UniversityChongqing 400044 the College of Information Science and Engineering Hunan UniversityChangsha 410082

Several recent successes in deep learning(DL),such as state-of-the-art performance on several image classification benchmarks,have been achieved through the improved ***(HPs)tuning is a key factor affecting the performance of machine learning(ML)*** state-of-the-art DL models use different HPs in different ways for classification tasks on different *** manuscript provides a brief overview of learning parameters and configuration techniques to show the benefits of using a large-scale handdrawn sketch dataset for classification *** analyzed the impact of different learning parameters and toplayer configurations with batch normalization(BN)and dropouts on the performance of the pre-trained visual geometry group 19(VGG-19).The analyzed learning parameters include different learning rates and momentum values of two different optimizers,such as stochastic gradient descent(SGD)and *** analysis demonstrates that using the SGD optimizer and learning parameters,such as small learning rates with high values of momentum,along with both BN and dropouts in top layers,has a good impact on the sketch image classification accuracy.

关键词： Deep learning(DL) hand-drawn sketches learning parameters

来源：评论

学校读者我要写书评

暂无评论

Artificial neural network with different learning parameters for crop classification using multispectral datasets

Artificial neural network with different learning parameters...

引用

International Conference on Microwave, Optical and Communication Engineering (ICMOCE)

作者： Kumar, Pradeep Prasad, Rajendra Mishra, Varun Narayan Gupta, Dileep Kumar Choudhary, Arti Srivastava, Prashant K. Indian Inst Technol BHU Dept Phys Varanasi Uttar Pradesh India Indian Inst Technol Dept Civil Engn Gauhati India Inst Environm & Sustainable Dev BHU Varanasi Uttar Pradesh India

ISBN: (纸本)9781467369817

Present study evaluated the performance of artificial neural network (ANN) algorithm using different learning parameters for various crop classification in Varanasi, India. Satellite images such as Linear Imaging Self Scanning (LISS) IV and Landsat 8 Operational Land Imager (OLI) were used for crop classification and comparative analysis study. The following crop such as barley, wheat, lentil, mustard, pigeon pea, linseed, corn, pea, sugarcane and other crops and non-crop such as water, sand, built up, fallow land, sparse vegetation and dense vegetation were identified in the area and classified. Results indicated a better classification accuracy of ANN algorithm for crop classification study when used with LISS-IV data in the comparison to Landsat 8-OLI multispectral satellite data. The larger values of the learning rates resulted high fluctuations and less classification accuracy using LISS-IV data, while less but nearly uniform results were found using the Landsat 8-OLI data.

关键词： LISS-IV Landsat 8-OLI learning parameters ANN classification accuracy

来源：评论

学校读者我要写书评

暂无评论

Distribution Matching for Machine Teaching

引用

IEEE TRANSACTIONS ON NEURAL NETWORKS AND learning SYSTEMS 2024年第9期35卷 12316-12329页

作者： Cao, Xiaofeng Tsang, Ivor W. Jilin Univ Sch Artificial Intelligence Changchun 130012 Peoples R China Minist Educ Engn Res Ctr Knowledge Driven Human Machine Intell Changchun 130012 Peoples R China Agcy Sci Technol & Res Ctr Frontier AI Res Singapore 138632 Singapore

Machine teaching is an inverse problem of machine learning that aims at steering the student toward its target hypothesis, in which the teacher has already known the student's learning parameters. Previous studies on machine teaching focused on balancing the teaching risk and cost to find the best teaching examples deriving from the student model. This optimization solver is in general ineffective when the student does not disclose any cue of the learning parameters. To supervise such a teaching scenario, this article presents a distribution matching-based machine teaching strategy via iteratively shrinking the teaching cost in a smooth surrogate, which eliminates boundary perturbations from the version space. Technically, our strategy could be redefined as a cost-controlled optimization process that finds the optimal teaching examples without further exploring the parameter distribution of the student. Then, given any limited teaching cost, the training examples would have a closed-form expression. Theoretical analysis and experiment results demonstrate the effectiveness of this strategy.

关键词： Education Training Optimization Machine learning Costs Task analysis Closed box learning parameters machine teaching surrogate teaching cost teaching risk

来源：评论

学校读者我要写书评

暂无评论

Adaptive Robot Behavior Based on Human Comfort Using Reinforcement learning

引用

IEEE ACCESS 2024年 12卷 122289-122299页

作者： Gonzalez-Santocildes, Asier Vazquez, Juan-Ignacio Eguiluz, Andoni Univ Deusto Fac Engn Bilbao 48007 Spain

This study explores the potential of training robots using reinforcement learning (RL) to adapt their behavior based on human comfort levels during tasks. An experimental environment has been developed and made available to the research community, facilitating the replication of these experiments. The results demonstrate that adjusting a single comfort-related input parameter during training leads to significant variations in the robot's behavior. Detailed discussions of the reward functions and obtained results validate these behavioral adaptations, confirming that robots can dynamically respond to human needs, thereby enhancing human-robot interaction. While the study highlights the effectiveness of this approach, it also raises the question of real-time comfort measurement, suggesting various systems for future exploration. These findings contribute to the development of more intuitive and emotionally responsive robots, offering new possibilities for future research in advancing human-robot interaction.

关键词： Robots Reinforcement learning Training Service robots Collaborative robots Human-robot interaction Behavioral sciences User experience Community environment human-robot interaction learning parameters reinforcement learning robot behavior task adaptation user comfort

来源：评论

学校读者我要写书评

暂无评论

A new learning space-variant anisotropic constrained-PDE for image denoising

引用

APPLIED MATHEMATICAL MODELLING 2024年第PartB期125卷 139-163页

作者： Hadri, Aissam Laghrib, Amine El Mourabit, Idriss Ibnou Zohr Univ Lab SIV Agadir Morocco Sultan Moulay Slimane Univ EMI FST Beni Mellal Beni Mellal Morocco Univ Cadi Ayyad EST Essaouira MIMSC Marrakech Morocco

In this paper, we propose an improved enhancement space-variant anisotropic PDE-constrained for image denoising, based on a learning optimization procedure. Since the tensor structure of the Weickert-type operators encodes three critical parameters: A1, A2 and 0, which define the local direction geometry in the image and also controls smoothing intensity along image features. We have then adopted an automatic estimation of these parameters based on a PDE-constrained optimization with a learning step of the additional information about the clean image. For the numerical solution of the computed free-noise image, we introduce a non-smooth Alternating Direction Method of Multipliers (ADMM) algorithm. The numerous tests in the experiments part incense, visually and quantitatively, the incomes of this new spatial tensor for PDE smoothing over the other denoising approaches.

关键词： Image denoising Spatial PDE ADMM learning parameters Tensor diffusion

来源：评论

学校读者我要写书评

暂无评论

Using the Kohonen Clustering Algorithm for Rough Peak Detection during Mass Spectrum Preprocessing

引用

JOURNAL OF ANALYTICAL CHEMISTRY 2023年第13期78卷 1800-1810页

作者： Lebedev, V. V. Buryak, A. K. Russian Acad Sci Frumkin Inst Phys Chem & Electrochem Moscow 119071 Russia

The paper explores the applicability of the Kohonen clustering algorithm for peak detection during the preprocessing of mass spectra. Various modifications made to the algorithm's classical version during its implementation in the program code are described . The results of peak detection carried out in four mass spectra of inorganic compounds using the Kohonen algorithm are presented. A brief study is conducted on how the algorithm's output is affected by various parameter values. The Kohonen algorithm and two other approaches are compared in terms of their peak detection quality. The findings suggest that the Kohonen clustering algorithm can be used for the initial detection of mass spectrum peaks;however, certain peculiarities of the algorithm's functioning need to be considered.

关键词： mass spectrometry peak detection Kohonen clustering algorithm machine learning learning parameters

来源：评论

学校读者我要写书评

暂无评论

A deep q-learning-based optimization of the inventory control in a linear process chain

引用

PRODUCTION ENGINEERING-RESEARCH AND DEVELOPMENT 2021年第1期15卷 35-43页

作者： Dittrich, M. -A. Fohlmeister, S. Inst Prod Engn & Machine Tools Univ 2 D-30823 Garbsen Germany

Due to growing globalized markets and the resulting globalization of production networks across different companies, inventory and order optimization is becoming increasingly important in the context of process chains. Thus, an adaptive and continuously self-optimizing inventory control on a global level is necessary to overcome the resulting challenges. Advances in sensor and communication technology allow companies to realize a global data exchange to achieve a holistic inventory control. Based on deep q-learning, a method for a self-optimizing inventory control is developed. Here, the decision process is based on an artificial neural network. Its input is modeled as a state vector that describes the current stocks and orders within the process chain. The output represents a control vector that controls orders for each individual station. Furthermore, a reward function, which is based on the resulting storage and late order costs, is implemented for simulations-based decision optimization. One of the main challenges of implementing deep q-learning is the hyperparameter optimization for the training process, which is investigated in this paper. The results show a significant sensitivity for the leaning rate alpha and the exploration rate epsilon. Based on optimized hyperparameters, the potential of the developed methodology could be shown by significantly reducing the total costs compared to the initial state and by achieving stable control behavior for a process chain containing up to 10 stations.

关键词： Inventory control Deep q-learning Process chain Self-optimizing control learning parameters

来源：评论

学校读者我要写书评

暂无评论

A New Approach Based on Fuzzy-Adaptive Structure & Parameter learning Applied in Meta-Cognitive Algorithm for ANFIS 7

A New Approach Based on Fuzzy-Adaptive Structure & Parameter...

引用

7th International Conference on Control, Instrumentation and Automation (ICCIA)

作者： Tavakoli, Parisa Vaezi, Nima Fahim, Parastou Karimpour, Ali Ferdowsi Univ Mashhad Dept Elect Engn Mashhad Razavi Khorasan Iran

ISBN: (纸本)9781665403504

A fuzzy inference system (FIS) is a widely used modeling and simulating approach based on the concept of fuzzy set theory, fuzzy if-then rules and fuzzy logic systems. All fuzzy inference systems have an adaptive composition that parameters are set during the training process. There is not any especial instruction to change the training parameters or the structure of the fuzzy inference systems during the training process. In this paper, a new online approach is proposed to adapt some systematic parameters of a typical fuzzy inference system. Employing this approach, some training and systematic parameters are selected during the training procedure to be constant or to be adaptive. This approach is realized and examined on the Adaptive Neural-fuzzy Inference System (ANFIS) and Meta-cognitive neuro fuzzy inference system (McFIS). The training process of the McFIS consists of two portions which includes a cognitive and a meta-cognitive one. A training mechanism controls the learning process of the cognitive portion. This paper illustrates the meta-cognitive training algorithm for parameterization of the fuzzy inference systems and applies the proposed approach for functional and systematic or structural parameter selection in the meta-cognitive algorithm (McFIS) which ends to some noticeable merits. The idea is examined by numerical examples and the results are compared with other methods. It is shown that the proposed approach reveals acceptable and efficient performance in the structure selection problem.

关键词： Fuzzy Adaptive Fuzzy Inference System learning parameters

来源：评论

学校读者我要写书评

暂无评论

Design of highly effective multilayer feedforward neural network by using genetic algorithm

引用

EXPERT SYSTEMS 2020年第4期37卷 e12532-e12532页

作者： Ustun, Oguz Bekiroglu, Erdal Onder, Mithat Abant Izzet Baysal Univ Fac Engn Dept Elect & Elect Engn TR-14280 Bolu Turkey

This paper presents a highly effective and precise neural network method for choosing the activation functions (AFs) and tuning the learning parameters (LPs) of a multilayer feedforward neural network by using a genetic algorithm (GA). The performance of the neural network mainly depends on the learning algorithms and the network structure. The backpropagation learning algorithm is used for tuning the network connection weights, and the LPs are obtained by the GA to provide both fast and reliable learning. Also, the AFs of each neuron in the network are automatically chosen by a GA. The present study consists of 10 different functions to accomplish a better convergence of the desired input-output mapping. Test studies are performed to solve a set of two-dimensional regression problems for the proposed genetic-based neural network (GNN) and conventional neural network having sigmoid AFs and constant learning parameters. The proposed GNN has also been tested by applying it to three real problems in the fields of environment, medicine, and economics. Obtained results prove that the proposed GNN is more effective and reliable when compared with the classical neural network structure.

关键词： activation functions genetic algorithm learning parameters multilayer feedforward neural networks

来源：评论

学校读者我要写书评

暂无评论

Selective Ensemble Least Square Support Vector Machine with Its Application

引用

IFAC-PapersOnLine 2018年第18期51卷 631-636页

作者： Tang, J. Qiao, J.F. Liu, Z. Wu, Z.W. Zhou, X.J. Yu, G. Zhao, J.J. Faculty of Information Technology Beijing University of Technology Beijing China State Key Laboratory of Synthetical Automation for Process Industries Northeaster University Shenyang110004 China State Key Laboratory of Process Automation in Mining & Metallurgy Beijing China Beijing Key Laboratory of Process Automation in Mining & Metallurgy Beijing China

Kernel-based modeling methods have been used widely to estimate some difficulty-to-measure quality or efficient indices at different industrial applications. Least square support vector machine (LSSVM) is one of the popular ones. However, its learning parameters, i.e., kernel parameter and regularization parameter, are sensitive to the training data and the model's prediction performance. Ensemble modeling method can improve the generalization performance and reliability of the soft measuring model. Aim at these problems, a new adaptive selective ensemble (SEN) LSSVM (SEN-LSSVM) algorithm is proposed by using multiple learning parameters. Candidate regularization parameters and candidate kernel parameters are used to construct many of candidate sub-sub-models based on LSSVM. These sub-sub-models based on the same kernel parameter are selected and combined as candidate SEN-sub-models by using branch and bound-based SEN (BBSEN). By employing BBSEN at the second time, these SEN-sub-models based on different kernel parameters are used to obtain the final soft measuring model. Thus, multiple kernel and regularization parameters are adaptive selected for building SEN-LSSVM model. UCI benchmark datasets and mechanical frequency spectral data are used to validate the effectiveness of this method. © 2018

关键词： learning systems Parameterization Support vector machines Generalization performance learning parameters Least square support vector machines Mechanical frequency Prediction performance Regularization parameters Selective ensemble modeling soft measuring

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：