检索结果-内蒙古大学图书馆

An explainable artificial intelligence model for multiple lung diseases classification from chest X-ray images using fine-tuned transfer learning

引用

Decision Analytics Journal 2024年 12卷

作者： Mahamud, Eram Fahad, Nafiz Assaduzzaman, Md Zain, S.M. Goh, Kah Ong Michael Morol, Md. Kishor Department of CSE Daffodil International University Faculty of Information Science and Technology Multimedia University Jalan Ayer Keroh Lama Department of Computing and Information Science Cornell University EliteLab.Ai

Traditional deep learning models are often considered “black boxes” due to their lack of interpretability, which limits their therapeutic use despite their success in classification tasks. This study aims to improve the interpretability of diagnoses for COVID-19, pneumonia, and tuberculosis from X-ray images using an enhanced DenseNet201 model within a transfer learning framework. We incorporated Explainable Artificial Intelligence (Xai) techniques, including SHAP, LIME, Grad-CAM, and Grad-CAM++, to make the model's decisions more understandable. To enhance image clarity and detail, we applied preprocessing methods such as Denoising Autoencoder, Contrast Limited Adaptive Histogram Equalization (CLAHE), and Gamma Correction. An ablation study was conducted to identify the optimal parameters for the proposed approach. Our model's performance was compared with other transfer learning-based models like EfficientNetB0, InceptionV3, and LeNet using evaluation metrics. The model that included data augmentation techniques achieved the best results, with an accuracy of 99.20%, and precision and recall of 99%. This demonstrates the critical role of data augmentation in improving model performance. SHAP and LIME provided significant insights into the model's decision-making process, while Grad-CAM and Grad-CAM++ highlighted specific image features and regions influencing the model's classifications. These techniques enhanced transparency and trust in ai-assisted diagnoses. Finally, we developed an Android-based system using the most effective model to support medical specialists in their decision-making process. © 2024 The Authors

关键词： Data augmentation Explainable artificial intelligence (Xai) Respiratory illnesses Transfer learning X-ray diagnosis

来源：评论

学校读者我要写书评

暂无评论

PCA-ViT: Hyperspectral Image Classification using Principal Component Analysis and Vision Transformer

PCA-ViT: Hyperspectral Image Classification using Principal ...

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Kian Ming Lim Chin Poo Lee Zaim Zahisham Jit Yan Lim Jashila Nair Mogan School of Computer Science University of Nottingham Ningbo China Zhejiang Province China School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Jiangsu Province China Faculty of Info. Science and Technology Multimedia University Melaka Malaysia Faculty of Info. Science & Technology Multimedia University Melaka Malaysia

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

Hyperspectral image classification is essential in remote sensing applications, aiming to accurately categorize land cover or materials depicted in hyperspectral data. This paper introduces PCA-ViT, a novel approach for hyperspectral image classification that integrates Principal Component Analysis (PCA) with the Vision Transformer (ViT) architecture. PCA serves as a spectral dimension reduction technique to alleviate the curse of dimensionality inherent in hyperspectral data, transforming the original data into a more manageable format. The transformed data are then input into the ViT model, which utilizes self-attention mechanisms to capture spatial dependencies among image patches, avoiding traditional convolutional layers. Extensive experiments on benchmark hyperspectral datasets, including Indian Pines, University of Pavia, and Salinas Scene, demonstrate PCA-ViT’s superior performance. It achieves $\mathbf{9 9. 9 5 \%}$ accuracy on Indian Pines, $\mathbf{1 0 0 \%}$ on University of Pavia, and 100% on Salinas Scene, showcasing the effectiveness of transformer-based architectures in hyperspectral image classification tasks.

关键词： Dimensionality reduction Computer vision Image processing Land surface Benchmark testing Transformers Data models Hyperspectral imaging Image classification Principal component analysis

来源：评论

学校读者我要写书评

暂无评论

Class Imbalance Pill Image Classification via Ensemble Learning

Class Imbalance Pill Image Classification via Ensemble Learn...

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Hui-Zong Chee Jit-Yan Lim Yong-Xuan Tan Kian-Ming Lim Chin-Poo Lee Pa-Pa-Min Faculty of Information Science & Technology (FIST) Multimedia University Melaka Malaysia School of Computer Science University of Nottingham Ningbo China Ningbo China School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Jiangsu Province China

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

Accurate identification of pill shapes is crucial for medication management and safety, acting as a safeguard against medication errors, especially with the increasing variety of pills that pose risks to patients. To address this, we propose a novel model for class imbalance pill image classification with the use of transfer learning and ensemble learning. Our proposed ensemble model integrates custom built Convolutional Neural Networks (CNN) and pre-trained EfficientNet-B2, both of which excel in image classification tasks. The CNN architecture employs multiple convolutional and pooling layers to identify pill characteristics, while EfficientNet enhances performance through its pre-trained capabilities. The proposed ensemble model demonstrated exceptional performance, achieving accuracies of 99.40% on the NIH Pill Image Dataset and 100% on the PillBR Dataset, respectively.

关键词： Accuracy Shape Transfer learning Safety Convolutional neural networks Ensemble learning Image classification Biomedical imaging

来源：评论

学校读者我要写书评

暂无评论

Brain Tumor Detection via Transfer Learning and Ensemble Learning

Brain Tumor Detection via Transfer Learning and Ensemble Lea...

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Wei-Jie Hun Jit-Yan Lim Kian-Ming Lim Chin-Poo Lee Yong-Xuan Tan Pa-Pa-Min Faculty of Information Science & Technology (FIST) Multimedia University Melaka Malaysia School of Computer Science University of Nottingham Ningbo China Ningbo China School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Jiangsu Province China

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

Brain tumors present a significant challenge in medical diagnostics due to their complexity and variability. This research investigates the application of deep learning to classify different types of brain tumors, aiming to enhance diagnostic accuracy and efficiency. The proposed ensemble method leverages a hybrid model that combines fine-tuned pre-trained DenseNet121 and EfficientNet-B0 models, omitting their respective top layers. By integrating the strengths of both networks through additional fully connected layers, classification performance is significantly improved. The proposed ensemble method achieves an accuracy of ${9 9. 6 7 \%}$ on the Figshare Dataset-I and ${9 9. 3 9 \%}$ on the Kaggle Dataset-II.

关键词： Deep learning Accuracy Transfer learning Brain tumors Brain modeling Robustness Complexity theory Ensemble learning Medical diagnosis

来源：评论

学校读者我要写书评

暂无评论

Transfer Learning-based Fruit Image Recognition

Transfer Learning-based Fruit Image Recognition

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Ting-Feng Low Isaac Jit-Yan Lim Yong-Xuan Tan Kian-Ming Lim Chin-Poo Lee Pa-Pa-Min Faculty of Information Science & Technology (FIST) Multimedia University Melaka Malaysia School of Computer Science University of Nottingham Ningbo China Ningbo China School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Jiangsu Province China

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

This research explores fruit image recognition using deep learning techniques, aiming to develop a robust and efficient model for accurately classifying diverse fruits. By leveraging transfer learning, the study applies pre-trained models with a particular focus on the EfficientNetV2-S architecture. The Fruits 360 dataset serves as the benchmark for evaluating the model’s performance. The research follows a systematic evaluation protocol, with accuracy as the key metric. Techniques such as fine-tuning are employed to enhance performance, and the study also considers the impact of model efficiency. The proposed method, based on the EfficientNetV2-S model, strikes a balance between accuracy and computational efficiency, providing valuable insights into the practical applications of deep learning for fruit image recognition.

关键词： Deep learning Training Adaptation models Image recognition Accuracy Systematics Protocols Computational modeling Transfer learning Data models

来源：评论

学校读者我要写书评

暂无评论

Cluster Boost: Stacked Machine Learning Model for Customer Behaviour Analysis in E-commerce

Cluster Boost: Stacked Machine Learning Model for Customer B...

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Zheng-Jie Wong Jit-Yan Lim Kian-Ming Lim Chin-Poo Lee Yong-Xuan Tan Ee-Mae Ang Faculty of Information Science & Technology (FIST) Multimedia University Melaka Malaysia School of Computer Science University of Nottingham Ningbo China Ningbo China School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Jiangsu Province China

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

This study focuses on the development of a stacked model, named Cluster Boost, which integrates K-means clustering and Gradient Boosting to analyse customer behaviour in e-commerce. Cluster Boost harnesses the strengths of both unsupervised and supervised learning methods, combining them to create a more robust and accurate predictive model. The proposed Cluster Boost model achieves an optimal testing accuracy of $\mathbf{9 7. 9 2 \%}$, significantly outperforming individual models such as Random Forest and Decision Tree on a public ecommerce customer churn analysis dataset. It also demonstrates an impressive precision of $\mathbf{9 5. 3 7 \%}$, a recall of $\mathbf{9 1. 9 6 \%}$, and an F 1 -score of $\mathbf{9 3. 6 4 \%}$. These results underscore the model’s potential for accurately predicting and understanding customer behaviour in e-commerce, highlighting the advantages of stacking K-means clustering with Gradient Boosting to capture intricate data relationships and enhance predictive accuracy.

关键词： Analytical models Accuracy Stacking Predictive models Boosting Electronic commerce Churn Predictive analytics Random forests Testing

来源：评论

学校读者我要写书评

暂无评论

Transfer Learning for Few-shot Pill Image Recognition

Transfer Learning for Few-shot Pill Image Recognition

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Hui-Zong Chee Jit-Yan Lim Kian-Ming Lim Chin-Poo Lee Yong-Xuan Tan Pa-Pa-Min Faculty of Information Science & Technology (FIST) Multimedia University Melaka Malaysia School of Computer Science University of Nottingham Ningbo China Ningbo China School of AI and Advanced Computing Xi’an Jiaotong-Liverpool University Jiangsu Province China

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

This study delves into few-shot learning techniques to enhance medication management and safety with limited pill image samples. Due to the limited data availability, which creates obstacles in achieving model generalization for conventional deep learning approaches, few-shot learning is adopted to handle this situation. In this work, we utilize a 5 -way 1 -shot learning setup with Prototypical Networks and ResNet-101 as the backbone to improve model generalization. Furthermore, transfer learning is incorporated to provide a strong knowledge foundation for the model to observe useful features from the limited training samples. The performance of the proposed model is evaluated on the CURE pill dataset under the 5 -way 1 -shot setting. The proposed model achieved an impressive accuracy of 89.65%, significantly outperforming existing methods.

关键词： Training Accuracy Image recognition Transfer learning Data models Safety Few shot learning Biomedical imaging Pharmaceuticals Image classification

来源：评论

学校读者我要写书评

暂无评论

Question Answering with Language Models

Question Answering with Language Models

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Jia Yin Wong Chin Poo Lee Kian Ming Lim Jit Yan Lim Jashila Nair Mogan Faculty of Info.Science & Technology Multimedia University Melaka Malaysia School of AI and Advanced Computing Xi’an Jiaotong Liverpool University Jiangsu Province China School of Computer Science University of Nottingham Ningbo China Zhejiang Province China

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

With the significant increase in the number of resources providing various types of information, ranging from thousands to millions of sources including research papers, blogs, and more, extracting and providing meaningful insights has become a challenge. This study develops a Question Answering (QA) system using the Haystack framework, capable of retrieving relevant documents and extracting answers from lengthy texts. The implemented QA system consists of four main components: an indexing pipeline, a document store, a searching pipeline, and an evaluation module. Additionally, this study investigates how different model architectures affect performance in retrieving and extracting answers. Two different retrievers, BM25 Retriever and Dense Passage Retriever, and five different readers, BERT, Roberta, albert, MiniLM, and ELECTRA, are examined. The models are tested and evaluated using the SQuAD datasets. The combination of BM25 Retriever and RoBERTa Reader achieved the best performance, with an F1-score of 0.9301 and an Exact Match score of $\mathbf{0. 8 9 5 6}$ on the SQuAD2.0 dataset.

关键词： Hands Analytical models System performance Pipelines Refining Process control Probabilistic logic Question answering (information retrieval) Distance measurement Indexing

来源：评论

学校读者我要写书评

暂无评论

Group re-identification with multi-grained matching and integration

arXiv

引用

arXiv 2019年

作者： Lin, Weiyao Li, Yuxi Xiao, Hao See, John Zou, Junni Xiong, Hongkai Wang, Jingdong Mei, Tao Department of Electronic Engineering Shanghai Jiao Tong University China Faculty of Computing and Informatics Multimedia University Malaysia Department of Computer Science Shanghai Jiao Tong University China Microsoft Research Asia Beijing China JD AI Research Beijing China

The task of re-identifying groups of people under different camera views is an important yet less-studied problem. Group re-identification (Re-ID) is a very challenging task since it is not only adversely affected by common issues in traditional single object Re-ID problems such as viewpoint and human pose variations, but it also suffers from changes in group layout and group membership. In this paper, we propose a novel concept of group granularity by characterizing a group image by multi-grained objects: individual persons and sub-groups of two and three people within a group. To achieve robust group Re-ID, we first introduce multi-grained representations which can be extracted via the development of two separate schemes, i.e. one with hand-crafted descriptors and another with deep neural networks. The proposed representation seeks to characterize both appearance and spatial relations of multi-grained objects, and is further equipped with importance weights which capture variations in intra-group dynamics. Optimal group-wise matching is facilitated by a multi-order matching process which in turn, dynamically updates the importance weights in iterative fashion. We evaluated on three multi-camera group datasets containing complex scenarios and large dynamics, with experimental results demonstrating the effectiveness of our approach. Copyright © 2019, The Authors. All rights reserved.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Acoustic Event Classification with Enhanced EfficientNet

Acoustic Event Classification with Enhanced EfficientNet

引用

IEEE Conference on Systems, Process & Control (ICSPC)

作者： Kian Ming Lim Chin Poo Lee Zhi Yang Lee Jit Yan Lim Jashila Nair Mogan School of Computer Science University of Nottingham Ningbo China Zhejiang Province China School of AI and Advanced Computing Xi’an Jiaotong Liverpool University Jiangsu Province China DZH International Sdn. Bhd. Kuala Lumpur Malaysia Faculty of Info. Science & Technology Multimedia University Melaka Malaysia

ISBN: (数字)9798350391398

ISBN: (纸本)9798350391404

In recent years, research into automating the recognition and classification of diverse acoustic events in audio recordings has surged. This technological advancement has profound implications for fields such as speech recognition, music information retrieval, and environmental sound monitoring. This study introduces a novel approach to acoustic event classification using a fine-tuned EfficientNet-B0 model. To mitigate overfitting, data augmentation techniques including pitch shifting, time stretching, noise addition, and time shifting are employed, thereby expanding the training dataset. Subsequently, these augmented audio signals undergo Short-Time Fourier Transform (STFT) to generate Log Mel-spectrograms, which are then integrated into the proposed fine-tuned EfficientNet-B0 architecture. Experimental results demonstrate promising performance across diverse settings, achieving validation accuracies of 89.44% and 74.23% on the ESC-10 and ESC- 50 datasets, respectively.

关键词： Training Accuracy Computational modeling Speech recognition Network architecture Data augmentation Acoustics Robustness Signal representation Spectrogram

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：