检索结果-内蒙古大学图书馆

Learning a Mixture of Conditional Gating Blocks for Visual Question Answering

Journal of computer science & technology 2024年第4期39卷 912-928页

作者： Qiang Sun Yan-Wei Fu Xiang-Yang Xue School of Statistics and Information Shanghai University of International Business and EconomicsShanghai 201620China Academy for Engineering and Technology Fudan UniversityShanghai 200433China School of Data Science Fudan UniversityShanghai 200433China School of Computer Science Fudan UniversityShanghai 200433China

As a Turing test in multimedia,visual question answering(VQA)aims to answer the textual question with a given ***,the“dynamic”property of neural networks has been explored as one of the most promising ways of improving the adaptability,interpretability,and capacity of the neural network ***,despite the prevalence of dynamic convolutional neural networks,it is relatively less touched and very nontrivial to exploit dynamics in the transformers of the VQA tasks through all the stages in an end-to-end ***,due to the large computation cost of transformers,researchers are inclined to only apply transformers on the extracted high-level visual features for downstream vision and language *** this end,we introduce a question-guided dynamic layer to the transformer as it can effectively increase the model capacity and require fewer transformer layers for the VQA *** particular,we name the dynamics in the Transformer as Conditional Multi-Head Self-Attention block(cMHSA).Furthermore,our questionguided cMHSA is compatible with conditional ResNeXt block(cResNeXt).Thus a novel model mixture of conditional gating blocks(McG)is proposed for VQA,which keeps the best of the Transformer,convolutional neural network(CNN),and dynamic *** pure conditional gating CNN model and the conditional gating Transformer model can be viewed as special examples of *** quantitatively and qualitatively evaluate McG on the CLEVR and VQA-Abstract *** experiments show that McG has achieved the state-of-the-art performance on these benchmark datasets.

关键词： visual question answering Transformer dynamic network

来源：评论

学校读者我要写书评

暂无评论

Normalized category travel personality by considering explicit and implicit feedback (NCTP): approach for improving travel recommender systems search result

引用

International Journal of Information technology (Singapore) 2023年第7期15卷 3689-3708页

作者： Kumar, Niranjan Hanji, Bhagyashri R. Department of Computer Science and Engineering Global Academy of Technology Visvesvaraya Technological University Belagavi 590018 India Department of Computer Science and Engineering Dayananda Sagar Academy of Technology and Management Visvesvaraya Technological University Belagavi 590018 India

Large volumes of end-user-generated textual data are assembled every day which leads to the evolution of social media in the form of reviews/feedback, and brief description messages. As a consequence, end-user often see it difficult to understand more concerning the subject being discussed or appropriate knowledge from such material. The enormous amount of text-oriented data that is accessible via the online platform is analyzed using machine learning and natural language processing algorithms, including topic modeling techniques that have been more prevalent in current years. The novel approach is proposed to represent travel categories called the Normalized Category Travel Personality (NCTP). The main purpose of the technique is to construct the semantics of category feedback to model travelers’ interests to create the category travel personality (CTP) representation. Likewise, we normalize the CTP to obtain our proposed model. The NCTP category model will apprehend the explicit and implicit aspects of categories shown by the social-circle groups to find sentiment scores. The TripAdvisor dataset was considered to evaluate the performance of the NCTP model based on the topic-model quality, implicit and explicit characteristics, and some legacy statistical evaluation metrics, like Recall, Mean Average Precision, and Mean Reciprocal Rank. © 2023, The Author(s), under exclusive licence to Bharati vidyapeeth's Institute of computer Applications and Management.

关键词： Probability distribution Rating and reviews Recommendation system Sentiment score

来源：评论

学校读者我要写书评

暂无评论

Enhancing Hospital Mortality Prediction for Heart Failure Patients through Hyperparameter Optimization and Interpretability Analysis with LIME and SHAP 1

Enhancing Hospital Mortality Prediction for Heart Failure Pa...

引用

1st International Conference for Women in Computing, InCoWoCo 2024

作者： Meena, S. Murthy, Monisha Kodipalli, Ashwini Global Academy of Technology Dept. of Computer Science and Engineering Karnataka Bengaluru India Global Academy of Technology Dept. of Artificial Intelligence & Data Science Karnataka Bengaluru India

ISBN: (纸本)9798331518943

Heart failure is one of the primary causes for deaths caused in the hospital. Predicting mortality rate of such patients is extremely important for the efficient use of health care resources. This research aims to estimate the hospital mortality rate of Heart Failure(HF) patients using machine learning(ML). Using computational technique, the obtained model has given the most efficient accuracy. This paper uses the classification model to detect two outcomes (0 and 1, i.e., if the patient will survive or not). The classification models used for the hospital mortality prediction are, single classifiers such as Logistic regression, SVM, KNN, Decision tree (Ginni and Entropy);with Logistic regression obtaining highest accuracy of 88.9%, Bagging and boosting models such as Randomforest, Adaptive boosting, Gradient boosting, XG boosting, Categorical boosting;with CatBoost obtaining highest accuracy of 90.2%. To achieve the best accuracy, ensemble models were used out of which ensemble classifier ensemble gave 91.1% accuracy. To understand the features playing important role in decision making, explainable AI(XAI) is used. LIME and SHAP are considered. LIME generated local explanations, it explained the decision of a complex model for a specific instance or observation. In contrast, SHAP provided global explanations, it explained the overall behavior of the model across all instances. © 2024 IEEE.

关键词： Adaptive boosting

来源：评论

学校读者我要写书评

暂无评论

Real-Time Facial Emotion Recognition Using Deep Learning Approach

Real-Time Facial Emotion Recognition Using Deep Learning App...

引用

2025 International Conference on Emerging Technologies for Intelligent Systems, ETIS 2025

作者： Chaitra, R. Vivekananda Bhat, K. Manipal Institute of Technology Manipal Academy of Higher Eucation Department of Computer Science and Engineering Manipal576104 India

ISBN: (数字)9798331507541

ISBN: (纸本)9798331507541

Classifying a person's emotional states is done using facial emotion recognition. The goal is to classify each face image into one of the 7 types of facial emotions: fear, disgust, surprise, sadness, neutral, happiness, and anger. To categorize the feeling CNN is utilized, and input is obtained via a variety of grayscale images via data collection and real-time videos. Subsequently, the CNN convolution and pooling layers are used for feature extraction, while the softmax layer is employed for categorization. Some techniques used to reduce the model's overfitting issue include dropout, cluster standardization, and L2 regularization. The model we developed outperforms previous efforts in accurately predicting individual emotions in the experiments conducted on the image collection of facial expressions. Additionally, the model performs well when used to forecast each image's sentiment using real-time video data. The developed deep learning model will collaborate with advancements in neuroscience, contributing to our understanding of the brain's mechanisms for emotion recognition. This may lead to more biologically inspired models and treatments for emotion-related disorders like autism. © 2025 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Skin Lesion Classification Using Feature Extraction and Ensemble Machine Learning Techniques 2

Skin Lesion Classification Using Feature Extraction and Ense...

引用

2nd International Conference on Machine Learning and Autonomous Systems, ICMLAS 2025

作者： Karnik, Arnav Sanjay Nair, Nikhil Dugar, Arham Narendra, V.G. Manipal Institute of Technology Manipal Academy of Higher Education Department of Computer Science and Engineering Manipal576104 India

ISBN: (纸本)9798331505745

This study explores a feature-engineering approach for classifying skin lesions as benign or malignant. Many other approaches regarding feature extraction can be applied: color, texture, shape, Gabor filters, Histogram of Oriented Gradients (HOG), edge density, fractal dimension, wavelet analysis, and entropy. It then gives a computationally efficient alternative instead of deep learning models. A soft-voting approach based on an ensemble of machine learning classifiers: SVM, MLP, and Random Forest is proposed. The accuracy of the classification task is significantly improved through the soft-voting approach. Experimental results exhibit the feasibility of the approach with an accuracy of 80 © 2025 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Detection of PCOS Leveraging Machine Learning and Interpretation Using Explainable AI 1

Detection of PCOS Leveraging Machine Learning and Interpreta...

引用

1st International Conference for Women in Computing, InCoWoCo 2024

作者： Madhavi, Bindu V. Firdouse, Asfiya Kodipalli, Ashwini Global Academy of Technology Dept. of Computer Science and Engineering Karnataka Bengaluru India Global Academy of Technology Dept. of Artificial Intelligence & Data Science Karnataka Bengaluru India

ISBN: (纸本)9798331518943

Polycystic Ovary Syndrome (PCOS) is a widespread endocrine disorder impacting women globally. This research aims to early predict and detect PCOS which is needed to reduce long-term complications. Since it is considered to be a hard-to-diagnose disorder, machine learning is utilized for this purpose. This study leveraged the power of computational algorithms trained on patient data for model prediction where Decision tree with criterion Gini index outperformed with 88.07% accuracy, 85.29% precision, 78.37% recall and 81.69% F1 Score. In addition to this, Bagging and boosting algorithms were used to monitor their performance metrics where Gradient Boost stood out with a remarkable accuracy of 91.74%, 91.00% precision, 97.00% recall and 94.00% F1 Score. By optimizing the chosen parameters through Hyperparameter tuning, a notable increase in the model's accuracy was observed. Besides, three ensemble models were proposed out of which the ensemble classifier ensemble model involving bagging, boosting and single classifiers brought a significant difference of 95.41% accuracy. Additionally, Explainable (XAI) methodologies like LIME (Local Interpretable Model-Agnostic Explanations) and SHAP (SHapley Additive exPlanations) were used for model interpretability. © 2024 IEEE.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Enhancing Financial Sentiment Analysis: Integrating the LoughranMcDonald Dictionary with BERT for Advanced Market Predictive Insights 3

Enhancing Financial Sentiment Analysis: Integrating the Loug...

引用

3rd International Conference on Machine Learning and Data engineering, ICMLDE 2024

作者： Sheetal, R. Aithal, Prakash K. Department of Computer Science & Engineering Manipal Institute of Technology Manipal Academy of Higher Education Manipal576104 India

One critical aspect of financial markets is understanding investor sentiment to facilitate effective decision-making. This study integrates traditional sentiment analysis methods, such as the Loughran-McDonald (LM) dictionary - designed for financial sentiment - with advanced Natural Language Processing (NLP) techniques using Bidirectional Encoder Representations from Transformers (BERT). The LM dictionary provides domain-specific word lists to label sentiment, whereas BERT enhances this by capturing nuanced meanings and semantic relationships in financial texts. It involves pre-processing Financial NewsHeadlines, applying the LM dictionary for sentiment scoring, and finetuning a pretrained BERT model to classify sentiment. A PyTorch dataset was created, tokenized using BERT, and processed through the model using techniques like dropout regularization and cross-entropy loss for optimization. The hybrid approach yields promising results: a classification accuracy of 97%, precision of 0.98, recall of 0.93, and an F1 score of 0.95, confirming its effectiveness in capturing sentiment polarity. In addition, comparisons between dictionary-labelled and pre-annotated datasets demonstrate the model's improved generalization ability. The results also show that our hybrid model outperformed various other existing models. This hybrid approach attempts to improve accuracy in capturing sentiment polarity by implementing methods to overcome imbalanced dataset, thereby facilitating a better understanding of sentiment in financial reports and facilitating informed decision-making. The integration of Named Entity Recognition (NER) with sentiment analysis based on sentiment polarity (positive, negative, or neutral) enables a more granular view of how specific companies are perceived in financial reports by highlighting the entities that are most affected by market sentiment. © 2024 The Authors. Published by ELSEVIER B.V.

关键词： Emotion Recognition

来源：评论

学校读者我要写书评

暂无评论

Discriminatively Constrained Semi-Supervised Multi-View Nonnegative Matrix Factorization with Graph Regularization

引用

Big Data Mining and Analytics 2024年第1期7卷 55-74页

作者： Guosheng Cui Ye Li Jianzhong Li Jianping Fan Shenzhen Institute of Advanced Technology Chinese Academy of SciencesShenzhen 518055China Joint Engineering Research Center for Health Big Data Intelligent Analysis Technology Shenzhen 518055China University of Chinese Academy of Sciences Beijing 100049China School of Computer Science and Control Engineering Shenzhen Institute of Advanced TechnologyChinese Academy of SciencesShenzhen 518055China

Nonnegative Matrix Factorization(NMF)is one of the most popular feature learning technologies in the field of machine learning and pattern *** has been widely used and studied in the multi-view clustering tasks because of its *** study proposes a general semi-supervised multi-view nonnegative matrix factorization *** algorithm incorporates discriminative and geometric information on data to learn a better-fused representation,and adopts a feature normalizing strategy to align the different *** specific implementations of this algorithm are developed to validate the effectiveness of the proposed framework:Graph regularization based Discriminatively Constrained Multi-View Nonnegative Matrix Factorization(GDCMVNMF)and Extended Multi-View Constrained Nonnegative Matrix Factorization(ExMVCNMF).The intrinsic connection between these two specific implementations is discussed,and the optimization based on multiply update rules is *** on six datasets show that the effectiveness of GDCMVNMF and ExMVCNMF outperforms several representative unsupervised and semi-supervised multi-view NMF approaches.

关键词： multi-view semi-supervised clustering discriminative information geometric information feature normalizing strategy

来源：评论

学校读者我要写书评

暂无评论

TwinNet: Twin Structured Knowledge Transfer Network for Weakly Supervised Action Localization

引用

Machine Intelligence Research 2022年第3期19卷 227-246页

作者： Xiao-Yu Zhang Hai-Chao Shi Chang-Sheng Li Li-Xin Duan Institute of Information Engineering Chinese Academy of SciencesBeijing 100093China School of Computer Science Beijing Institute of TechnologyBeijing 100081China School of Computer Science and Engineering University of Electronic Science and Technology of ChinaChengdu 611731China

Action recognition and localization in untrimmed videos is important for many applications and have attracted a lot of attention. Since full supervision with frame-level annotation places an overwhelming burden on manual labeling effort, learning with weak video-level supervision becomes a potential solution. In this paper, we propose a novel weakly supervised framework to recognize actions and locate the corresponding frames in untrimmed videos simultaneously. Considering that there are abundant trimmed videos publicly available and well-segmented with semantic descriptions, the instructive knowledge learned on trimmed videos can be fully leveraged to analyze untrimmed videos. We present an effective knowledge transfer strategy based on inter-class semantic relevance. We also take advantage of the self-attention mechanism to obtain a compact video representation, such that the influence of background frames can be effectively eliminated. A learning architecture is designed with twin networks for trimmed and untrimmed videos, to facilitate transferable self-attentive representation learning. Extensive experiments are conducted on three untrimmed benchmark datasets (i.e., THUMOS14, ActivityNet1.3, and MEXaction2), and the experimental results clearly corroborate the efficacy of our method. It is especially encouraging to see that the proposed weakly supervised method even achieves comparable results to some fully supervised methods.

关键词： Knowledge transfer weakly supervised learning self-attention mechanism representation learning action localization

来源：评论

学校读者我要写书评

暂无评论

Symptoms-based Classification of Disease Using Various ML Algorithms and Interpretation using LIME and SHAP KERNELS 4

Symptoms-based Classification of Disease Using Various ML Al...

引用

4th IEEE Asian Conference on Innovation in technology, ASIANCON 2024

作者： Soujanya, K.J. Kodipalli, Ashwini Gosai, Sanjeev Sneha, B.J. Rao, Trupthi Global Academy of Technology Dept. of Artificial Intelligence & Data Science Karnataka Bangalore India Global Academy of Technology Dept. of Computer Science & Engineering Karnataka Bangalore India

ISBN: (纸本)9798350354218

Using a variety of machine learning techniques, this research study suggests a unique method for classifying diseases using symptom-based analysis. To improve model transparency and comprehension, the study makes use of the interpretability tools, SHAP kernels and LIME. The research attempts to close the gap between interpretability and prediction accuracy in medical diagnosis by combining two approaches. A thorough assessment of symptom-based categorization is ensured by the use of many machine learning methods. The application of LIME and SHAP kernels aids in the clarification of model predictions and provides information on the primary symptoms that impact illness categorization. This study advances the creation of accurate and comprehensible illness categorization models, promoting confidence in the use of machine learning methods in healthcare. © 2024 IEEE.

关键词： Adversarial machine learning

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：