检索结果-内蒙古大学图书馆

Using a one-dimensional convolutional neural network on FTIR spectroscopy to measure the thickness of composite plastic films

引用

INFRARED PHYSICS & TECHNOLOGY 2025年 147卷

作者： Wang, Xiaodong Ni, Liwei Zhang, Cheng Xu, Qiyue Ye, Shuliang China Jiliang Univ Inst Thermal Anal Technol & Instrumentat 258 Xueyuan St Hangzhou 310018 Peoples R China

Multilayer composite films have gained widespread application across a variety of industries and applications due to their unique properties and functionalities. The precise control of the thickness of each layer ensures that every layer contributes optimally to the desired attributes. In this study, a method using a one-dimensional convolutional neural network (1D-CNN) was developed to analyze the spectral data, five kinds of plastic films were assembled with different thickness combinations and measured with FTIR, which provided the data sets for 1D-CNN modeling. Compared to partial least squares (PLS) and fully connected neural network (FCNN) models, the 1D-CNN model performs better in accuracy of the thickness prediction, with respective root mean square error (RMSE) values dropping from < 9.11 mu m to < 0.31 mu m for film thickness of 13-270 mu m. The developed regression model was visualized using the gradient weighted class activation mapping (Grad-CAM) method for detailed analysis and selection of significant wavenumbers in order to build a more compact 1D-CNN model. Altogether, these results show that the proposed spectral 1D-CNN technique can measure the thickness of components in composite plastic films rapidly and accurately, and has the potential to improve efficiency and reduce the cost of instruments and calculations for spectral quantitation applications.

关键词： Composite plastic film Infrared spectrum One-dimensional convolutional neural network neural network visualization

来源：评论

学校读者我要写书评

暂无评论

High-Impedance Fault Detection Methodology Using Time-Frequency Spectrum and Transfer Convolutional neural network in Distribution networks

引用

IEEE SYSTEMS JOURNAL 2023年第3期17卷 4002-4013页

作者： Guo, Mou-Fa Guo, Zi-Yi Gao, Jian-Hong Chen, Duan-Yu Fuzhou Univ Coll Elect Engn & Automat Fuzhou 350108 Peoples R China Yuan Ze Univ Dept Elect Engn Taoyuan 32003 Taiwan

High-impedance fault (HIF) detection has always been difficult in distribution networks due to the lack of field data and the large difference between field and simulation waveforms. Based on the characteristics of zero-sequence currents, a novel HIF detection methodology is proposed, which combines time-frequency spectrum (TFS) and transfer convolutional neural network (TCNN). First, the TFSs are acquired by applying continuous wavelet transform (CWT) to the collected zero-sequence currents. Then, the TFSs of simulated zero-sequence currents are utilized for training source-domain convolutional neural network (SCNN). Next, the SCNN is transfer learned with very few TFSs of field zero-sequence currents to obtain TCNN. The performance of the proposed method is verified by simulation samples and field samples. The results show that the proposed method can effectively extract fault features from small-scale training samples under different fault circumstances. Besides, TCNN can adaptively extract the effective features of field HIF and detect field HIF more accurately than SCNN. Finally, this article provides a visualization scheme for interpretability of the neural network, which offers visual explanations for the decision-making basis of the neural network.

关键词： Distribution networks high-impedance fault (HIF) neural network visualization transfer convolutional neural network (TCNN) transfer learning

来源：评论

学校读者我要写书评

暂无评论

Seismic impedance inversion based on cycle-consistent generative adversarial network

引用

Petroleum Science 2022年第1期19卷 147-161页

作者： Yu-Qing Wang Qi Wang Wen-Kai Lu Qiang Ge Xin-Fei Yan The Institute for Artificial Intelligence Tsinghua University(THUAI)Beijing100084China State Key Laboratory of Intelligent Technology and Systems Tsinghua UniversityBeijing100084China Beijing National Research Center for Information Science and Technology(BNRist) Tsinghua UniversityBeijing100084China The Department of Automation Tsinghua UniversityBeijing100084China The Research Institute of Petroleum Exploration and Development China National Petroleum Corporation(CNPC)Beijing100083China

Deep learning has achieved great success in a variety of research fields and industrial ***,when applied to seismic inversion,the shortage of labeled data severely influences the performance of deep learning-based *** order to tackle this problem,we propose a novel seismic impedance inversion method based on a cycle-consistent generative adversarial network(Cycle-GAN).The proposed Cycle-GAN model includes two generative subnets and two discriminative *** kinds of loss,including cycle-consistent loss,adversarial loss,and estimation loss,are adopted to guide the training *** from the proposed structure,the information contained in unlabeled data can be extracted,and adversarial learning further guarantees that the prediction results share similar distributions with the real ***,a neural network visualization method is adopted to show that the proposed CNN model can learn more distinguishable features than the conventional CNN *** robustness experiments on synthetic data sets show that the proposed method can achieve better performances than other methods in most *** the blind-well experiments on real seismic profiles show that the predicted impedance curve of the proposed method maintains a better correlation with the true impedance curve.

关键词： Seismic inversion Cycle GAN Deep learning Semi-supervised learning neural network visualization

来源：评论

学校读者我要写书评

暂无评论

SkipPLUS: Skip the First Few Layers to Better Explain Vision Transformers

SkipPLUS: Skip the First Few Layers to Better Explain Vision...

引用

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Mehri, Faridoun Fayyaz, Mohsen Baghshah, Mahdieh Soleymani Pilehvar, Mohammad Taher Sharif Univ Technol Tehran Iran Univ Tehran Tehran Iran Cardiff Univ Cardiff Wales

ISBN: (纸本)9798350365474

Despite their remarkable performance, the explainability of Vision Transformers (ViTs) remains a challenge. While forward attention-based token attribution techniques have become popular in text processing, their suitability for ViTs hasn't been extensively explored. In this paper, we compare these methods against state-of-the-art input attribution methods from the Vision literature, revealing their limitations due to improper aggregation of information across layers. To address this, we introduce two general techniques, PLUS and SkipPLUS, that can be composed with any input attribution method to more effectively aggregate information across layers while handling noisy layers. Through comprehensive and quantitative evaluations of faithfulness and human interpretability on a variety of ViT architectures and datasets, we demonstrate the effectiveness of PLUS and SkipPLUS, establishing a new state-of-the-art in white-box token attribution. We conclude with a comparative analysis highlighting the strengths and weaknesses of the best versions of all the studied methods. The code used in this paper is freely available at https://***/NightMachinery/SkipPLUS-CVPR-2024.

关键词： Explainable AI Forward Attention-Based Token Attribution Interpretability neural network visualization Vision Transformers White-Box Input Attribution Methods xAI

来源：评论

学校读者我要写书评

暂无评论

neural network training fingerprint: visual analytics of the training process in classification neural networks

引用

JOURNAL OF visualization 2022年第3期25卷 593-612页

作者： Ferreira, Martha Dais Cantareira, Gabriel D. de Mello, Rodrigo F. Paulovich, Fernando V. King's Coll London Dept Informatics London England Univ Sao Paulo Sao Carlos Brazil Dalhousie Univ Fac Comp Sci Halifax NS Canada Kings Coll London Dept Informat London England Univ Sao Paulo Sao Carlos Brazil

The striking results of deep neural networks (DNN) have motivated its wide acceptance to tackle large datasets and complex tasks such as natural language processing, facial recognition, and artificial image generation. However, DNN parameters are often empirically selected on a trial-and-error approach without detailed information on convergence behavior. While some visualization techniques have been proposed to aid the comprehension of general-purpose neural networks, only a few explore the training process, lacking the ability to adequately display how abstract representations are formed and represent the influence of training parameters during this process. This paper describes neural network training fingerprint (NNTF), a visual analytics approach to investigate the training process of any neural network performing classification. NNTF allows understanding how classification decisions change along the training process, displaying information about convergence, oscillations, and training rates. We show its usefulness through case studies and demonstrate how it can support the analysis of training parameters.

关键词： neural network visualization neural network training Deep learning Visual analytics visualization

来源：评论

学校读者我要写书评

暂无评论

Interpretable functional specialization emerges in deep convolutional networks trained on brain signals

引用

JOURNAL OF neural ENGINEERING 2022年第3期19卷 036006-036006页

作者： Hammer, J. Schirrmeister, R. T. Hartmann, K. Marusic, P. Schulze-Bonhage, A. Ball, T. Univ Freiburg Med Ctr Univ Freiburg Fac Med Dept Neurosurg Freiburg Germany Charles Univ Prague Fac Med 2 Dept Neurol Prague Czech Republic Motol Univ Hosp Prague Czech Republic Univ Freiburg Med Ctr Univ Freiburg Fac Med Epilepsy Ctr Freiburg Germany Univ Freiburg Fac Engn Dept Comp Sci Machine Learning Lab Freiburg Germany

Objective. Functional specialization is fundamental to neural information processing. Here, we study whether and how functional specialization emerges in artificial deep convolutional neural networks (CNNs) during a brain-computer interfacing (BCI) task. Approach. We trained CNNs to predict hand movement speed from intracranial electroencephalography (iEEG) and delineated how units across the different CNN hidden layers learned to represent the iEEG signal. Main results. We show that distinct, functionally interpretable neural populations emerged as a result of the training process. While some units became sensitive to either iEEG amplitude or phase, others showed bimodal behavior with significant sensitivity to both features. Pruning of highly sensitive units resulted in a steep drop of decoding accuracy not observed for pruning of less sensitive units, highlighting the functional relevance of the amplitude- and phase-specialized populations. Significance. We anticipate that emergent functional specialization as uncovered here will become a key concept in research towards interpretable deep learning for neuroscience and BCI applications.

关键词： motor decoding intracranial EEG (iEEG) deep learning brain-computer interface (BCI) neural network visualization internal representation explainable AI (XAI)

来源：评论

学校读者我要写书评

暂无评论

Deep Insights into Convolutional networks for Video Recognition

引用

INTERNATIONAL JOURNAL OF COMPUTER VISION 2020年第2期128卷 420-437页

作者： Feichtenhofer, Christoph Pinz, Axel Wildes, Richard P. Zisserman, Andrew Graz Univ Technol Graz Austria York Univ Toronto ON Canada Univ Oxford Oxford England

As the success of deep models has led to their deployment in all areas of computer vision, it is increasingly important to understand how these representations work and what they are capturing. In this paper, we shed light on deep spatiotemporal representations by visualizing the internal representation of models that have been trained to recognize actions in video. We visualize multiple two-stream architectures to show that local detectors for appearance and motion objects arise to form distributed representations for recognizing human actions. Key observations include the following. First, cross-stream fusion enables the learning of true spatiotemporal features rather than simply separate appearance and motion features. Second, the networks can learn local representations that are highly class specific, but also generic representations that can serve a range of classes. Third, throughout the hierarchy of the network, features become more abstract and show increasing invariance to aspects of the data that are unimportant to desired distinctions (e.g. motion patterns across various speeds). Fourth, visualizations can be used not only to shed light on learned representations, but also to reveal idiosyncrasies of training data and to explain failure cases of the system.

关键词： Computer vision Machine learning Deep learning Video recognition neural network visualization Action recognition

来源：评论

学校读者我要写书评

暂无评论

Visualizing Learning Space in neural network Hidden Layers 15

Visualizing Learning Space in Neural Network Hidden Layers

引用

15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP) / 11th International Conference on Information visualization Theory and Applications (IVAPP)

作者： Cantareira, Gabriel D. Paulovich, Fernando, V Etemad, Elham Univ Sao Paulo Sao Paulo Brazil Dalhousie Univ Halifax NS Canada

ISBN: (纸本)9789897584022

Analyzing and understanding how abstract representations of data are formed inside deep neural networks is a complex task. Among the different methods that have been developed to tackle this problem, multidimensional projection techniques have attained positive results in displaying the relationships between data instances, network layers or class features. However, these techniques are often static and lack a way to properly keep a stable space between observations and properly convey flow in such space. In this paper, we employ different dimensionality reduction techniques to create a visual space where the flow of information inside hidden layers can come to light. We discuss the application of each used tool and provide experiments that show how they can be combined to highlight new information about neural network optimization processes.

关键词： Machine Learning neural network visualization Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Exploring neural network Hidden Layer Activity Using Vector Fields

引用

INFORMATION 2020年第9期11卷 426页

作者： Cantareira, Gabriel D. Etemad, Elham Paulovich, Fernando V. Univ Sao Paulo Inst Ciencias Matemat & Computacao BR-13566590 Sao Paulo Brazil Dalhousie Univ Fac Comp Sci Halifax NS B3H 4R2 Canada

Deep neural networks are known for impressive results in a wide range of applications, being responsible for many advances in technology over the past few years. However, debugging and understanding neural networks models' inner workings is a complex task, as there are several parameters and variables involved in every decision. Multidimensional projection techniques have been successfully adopted to display neural network hidden layer outputs in an explainable manner, but comparing different outputs often means overlapping projections or observing them side-by-side, presenting hurdles for users in properly conveying data flow. In this paper, we introduce a novel approach for comparing projections obtained from multiple stages in a neural network model and visualizing differences in data perception. Changes among projections are transformed into trajectories that, in turn, generate vector fields used to represent the general flow of information. This representation can then be used to create layouts that highlight new information about abstract structures identified by neural networks.

关键词： explainable machine learning neural network visualization dimensionality reduction

来源：评论

学校读者我要写书评

暂无评论

An optimized convolutional neural network with bottleneck and spatial pyramid pooling layers for classification of foods

引用

PATTERN RECOGNITION LETTERS 2018年 105卷 50-58页

作者： Jahani Heravi, Elnaz Habibi Aghdam, Hamed Puig, Domenec Univ Rovira & Virgili Dept Comp Engn & Math Tarragona Spain

Keeping record of daily meal intake is an effective solution for tackling with obesity and overweight. This can be done by developing apps on smartphones that are able to automatically recommend a short list of most probable foods by analyzing the photo taken from food. Then, the user chooses the correct answer from the short list. Hence, the automatic food recognition system must be able to recommend an accurate list. In other words, it is not essential for these apps to have a very high top-1 accuracy. Considering that the app will show the list of 5 most probable foods, the food recognition system must have a high top-5 accuracy. A food recognition system is usually developed by adapting knowledge of state-of-the-art networks such as GoogleNet and ResNet to the domain of food. However, these networks have high number of parameters. In this paper, we propose a 23-layer architecture which has 99.14% and 96.63% fewer parameter compared with ResNet and GoogleNet. Our experiment on Food101 and UECFood-256 datasets shows that although our network reduces the number of parameters dramatically, it produces more accurate results than GoogleNet and its accuracy is comparable with ResNet. (c) 2017 Elsevier B.V. All rights reserved.

关键词： Food classification Convolutional neural networks neural network visualization Deep learning Spatial pyramid pooling

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

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

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：