检索结果-内蒙古大学图书馆

4th International Conference on Emerging Trends in Networks and Computer Communications, ETNCC 2024

作者： Njathi, Yuri Maina, Ciira Wa Dedan Kimathi University of Technology Centre for Data Science and Artificial Intelligence Nyeri Kenya

ISBN: (纸本)9798350353266

Traditionally classification is performed on originally collected images due to its computational simplicity. Preprocessing steps are usually conducted before classification. This research aims to examine what effects segmenting out the background as well as replacing the background with natural landscapes has on classification performance for out-of-the-box state-of-the-art models. This paper summarizes research in image classification and the impact of impala background on the performance of a zero-shot MobileNetv3 classification model utilizing ImageNet weights. Most of the research is experimental in nature and involves the utilization of DSAIL-Porini images and labels, an open-source wildlife image dataset. The methodology involved the input of impala images into a pretrained model for classification and assessing the model's performance. The input images were all single-class, impala images. Segment Anything Model (SAM) and You Only Look Once version 8 (YOLOv8) detection models were used to remove the image backgrounds with the impalas as the foreground and a similar background with undetectable objects was also used to access the model's performance on the third experiment. The results show that upon removal of the background, the model's accuracy dropped significantly from 18.15% to 0.11%. Upon swapping the background with a similar conservancy background, there was a less significant drop in performance, 2.05%. This implies a significant effect of the background on classification performance. © 2024 IEEE.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Default credit card scoring using ML 1

Default credit card scoring using ML

引用

1st International Conference on Computational intelligence for Security, Communication and Sustainable Development, CISCSD 2024

作者： Darshan, B.S. Vishal Kumar, I.B. Kodipalli, Ashwini Rao, Trupti Global Academy of Technology Department of Artificial Intelligence and Data Science Karnataka Bengaluru India

ISBN: (纸本)9798350365405

Default credit scoring is a critical process that estimates the probability of a person or organization failing to meet their credit obligations, such as not making a loan or Credit Card payment on time. This evaluation is vital for financial institutions as it informs their decision to grant credit to an applicant and sets the terms of the credit, including interest rates and credit limits. In essence, default credit scoring enables financial institutions to make informed lending decisions, reduce potential losses from defaults, and sustain a robust lending portfolio. It serves a crucial function in the financial industry by assessing the creditworthiness of individuals applying for credit cards. Historically, default credit scoring methods have relied on basic rule-based systems or quantitative frameworks, which often have limited predictive power. However, the rise of ML methods has evolved to this in recent years process, providing improved accuracy and efficiency. This article offers exhaustive feedback of the computational learning methods employed for default credit card scoring, such as logistic regression, decision trees, random forests, and support vector machines. It also discusses choosing attributes and pre-processing methods to enhance model performance and interpretability, along with strategies to address these issues. By comparing various ML approaches, this study provides a deep understanding of the latest methods for default credit card scoring and suggests directions for future research. © 2024 IEEE.

关键词： Forestry

来源：评论

学校读者我要写书评

暂无评论

Next Word Prediction System using NLP

Next Word Prediction System using NLP

引用

2024 International Conference on Electrical, Electronics and Computing Technologies, ICEECT 2024

作者： Shivnani, Hiya Sontakke, Rashmi Sisode, Bhushan Puri, Pranesh Patil, Ratna Vishwakarma Institute of Information Technology Department of Artificial Intelligence and Data Science Pune India

ISBN: (纸本)9798350378092

This research paper introduces a novel statistical trigram language model designed for text prediction tasks, implemented using the Natural Language Toolkit (NLTK). The model is trained on the renowned 'Alice in Wonderland' corpus, leveraging a weighted random probability approach to predict the subsequent word in a provided phrase. A distinctive feature of this model is its user-controlled prediction termination, allowing individuals to determine when the prediction process should halt. To ensure data quality, the text undergoes rigorous preprocessing stages, eliminating superfluous characters and ensuring standardized tokenization into trigrams through NLTK's robust capabilities. The research not only presents the predictive model but also includes comprehensive n-gram statistics, offering insights into the underlying linguistic structures. Additionally, a visually informative frequency distribution plot showcasing the top 30 most common trigrams enriches the analysis, aiding in understanding the corpus's textual patterns. This paper contributes to advancing text prediction methodologies by combining statistical modeling, NLTK proficiency, and user interactivity, offering a valuable tool for language enthusiasts, researchers, and developers alike. © 2024 IEEE.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Driver's Safety Measures: A Blind Spot and Drowsiness Detection System using IoT Components and Python Library 15

Driver's Safety Measures: A Blind Spot and Drowsiness Detect...

引用

15th International Conference on Advances in Computing, Control, and Telecommunication Technologies, ACT 2024

作者： Mundargi, Zarinabegam Ingale, Bhushan Jadhav, Makarand Kamble, Omkar Kayade, Parth Lanjewar, Harsh Department of Artificial Intelligence and Data Science Vishwakarma Institute of Technology Pune411037 India

ISBN: (纸本)9798331300579

This research paper examines the implementation of advanced driver safety measures focused on drowsiness and blind spot detection, utilizing innovative technologies including OpenCV, Firmata, and PyMata. As traffic accidents continue to be a significant concern worldwide, proactive measures to enhance driver safety are crucial. The study explores the development and integration of drowsiness detection systems and blind spot detection technologies in vehicles to mitigate the risk of accidents caused by these factors. This research paper provides a comprehensive overview of the current state of driver safety measures involving drowsiness and blind spot detection. It highlights the advancements, challenges, and opportunities in these areas, ultimately contributing to the development of safer transportation systems and reducing the risk of accidents caused by drowsiness and blind spot-related factors. © Grenze Scientific Society, 2024.

关键词： Safety factor

来源：评论

学校读者我要写书评

暂无评论

Partially-Supervised Graph Derivation Network with Meta-Learning for Time Series Anomaly Detection

引用

IEEE Internet of Things Journal 2025年第13期12卷 25472-25486页

作者： Zhu, Sanli Li, Yuan Xu, Kang Xu, Junjun College of Automation and Artificial Intelligence Nanjing University of Posts and Telecommunications Nanjing China Nanjing University of Science and Technology China School of Computer Science Nanjing University of Posts and Telecommunications Nanjing University of Science and Technology Nanjing China

Time series anomaly detection is essential in various fields such as industrial monitoring, cybersecurity, and finance. Traditional supervised methods often face challenges due to the limited availability of labeled anomaly instances for training. Moreover, these methods struggle to deal with intricate systems that incorporate information about topological structures. In this paper, we propose a novel approach called the Partially-Supervised Graph Derivation Network with Meta Learning (PS-GDNML) for time series anomaly detection. PS-GDNML combines the power of graph-based representations, partially-supervised learning, and meta-learning to enhance the effectiveness and robustness of anomaly detection. The method represents time series data as a graph, where each data point is a node, and temporal relationships are captured through edges. By leveraging a graph attention neural network (GAT), the model effectively captures complex dependencies and relationships within the data. To address the scarcity of labeled anomaly instances, PS-GDNML adopts a partially-supervised learning framework. It utilizes both labeled and unlabeled data, enabling the model to learn from the available information and generalize to detect anomalies in unseen data. Additionally, to explore the underlying commonalities of data across different time periods and enhance the model’s adaptability, we adopted a new meta-learning method called Task Relation Meta-Learner (TRMLearner). The purpose of this project is to utilize task relationships to guide the meta-learning optimization process. We evaluated the performance of PS-GDNML on benchmark datasets and compared it with state-of-the-art anomaly detection methods. The experimental results demonstrate that, even with a restricted set of labeled instances, our method excels at accurately detecting anomalies. Furthermore, the meta-learning component enhances the model’s capacity to generalize to novel and evolving anomaly patterns. © 2014 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Next Word Prediction System Using NLP 1

Next Word Prediction System Using NLP

引用

1st IEEE International Conference on Smart Power Control and Renewable Energy, ICSPCRE 2024

作者： Patil, Ratna Dolas, Vaishnavi Godse, Ritesh Waghmare, Soham Chopade, Aniket Vishwakarma Institute of Information Technology Department of Artificial Intelligence and Data Science Pune India

ISBN: (纸本)9798350377002

This study focuses on Natural Language Processing (NLP) by developing a Next Word Prediction system utilizing Long Short-Term Memory (LSTM) networks. The system is built to take advantage of LSTM's capacity to capture temporal dependencies in text data using TensorFlow's Keras package. The major objective is to enhance content development and writing aid by precisely predicting the following word in a series. This will facilitate various apps, including virtual assistants., and increase typing speed. The first portion of the paper examines several NLP techniques., with a focus on Recurrent Neural Networks (RNNs) and some of their variants., including LSTM. The effectiveness of different architectural configurations and training methods is investigated through comprehensive testing and review to maximize predicted accuracy. The literature review sheds light on current methods., such as hybrid models., inflected language processing strategies., and deep learning architectures like Transformers. The methodology is informed by these sources and consists of preprocessing the data., modeling with Sequential LSTM architecture, and carrying out a comprehensive evaluation with metrics such as top-k accuracy and perplexity. Additional validation for the suggested approach comes from comparative evaluations using ablation experiments and baseline models. The results show that the LSTM-based model's high accuracy over a variety of datasets indicates its capacity to understand intricate linguistic patterns. Analyzing intricate systems., adding context., and modifying models for certain domains are a few possible areas for development. The project's end highlights the importance of LSTM networks in language modeling. © 2024 IEEE.

关键词： Long short-term memory

来源：评论

学校读者我要写书评

暂无评论

Personalized Music Recommendation: A Comparative Study of Machine Learning and Deep Learning Approaches 11

Personalized Music Recommendation: A Comparative Study of Ma...

引用

11th IEEE International Conference on Advances in Computing and Communications, ICACC 2024

作者： Iqbal, Nafla Mohan, Aiswarya Syjo, Ann Rose Rajagiri School of Engineering and Technology Dept of Artificial Intelligence and Data Science Kakkanad India

ISBN: (纸本)9798350379136

The rapid advancement of music playback technology, particularly within mobile platforms, has fundamentally transformed user engagement with music. Despite significant improvements in music retrieval techniques over the past decade, the evolution of music recommendation systems remains in its early stages. With users having access to extensive online music libraries, the task of selecting preferred tracks from an overwhelming array of options presents a considerable challenge. Individual musical tastes vary widely, and the influx of new users and music necessitates a system that can adapt *** music recommendation systems face the dual challenge of managing vast amounts of data while delivering accurate results, making them resource-intensive. However, the high precision of these models drives their increasing application in commercial settings. This paper proposes a framework aimed at enhancing recommendation accuracy by tailoring suggestions to align more closely with user preferences.A comprehensive analysis of traditional and advanced music recommendation methodologies is presented, focusing on two user-centric approaches: user preference and emotion-based metrics. These methodologies are critically compared, leveraging the Million Song dataset (MSD) for training the recommendation system. Additionally, an Android application has been developed to collect empirical data on user experiences, facilitating a deeper understanding of user interactions and satisfaction with the recommendation framework. This research aims to improve user experience in music discovery by providing more relevant and personalized recommendations. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Machine Failure Detection Using artificial Neural Networks and a Comparison of Various Optimizers 4

Machine Failure Detection Using Artificial Neural Networks a...

引用

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

作者： Raj, Harini Kumar, Ashwin Kodipalli, Ashwini Rao, Trupthi Global Academy of Technology Department of Artificial Intelligence and Data Science Karnataka Bangalore India

ISBN: (纸本)9798350354218

Accurate identification of evolving failures in machines is an essential factor for industrial operations in a venture to reduce costs and disruptions affiliated to machine breakdowns. Machine failures not only compensate for the time lost waiting for the machine to be repaired but also result in costs that are incurred in repairing the damage and most times they lead to loss making instances. This research aims to construct a model of ANNs for predicting these failures. This research improves the current literature by comparing distinct optimization algorithms systematically to improve the ANN model's performance. The layers include numerous dense layers with Activation function like Relu, several dropout layers for reducing overfitting, and L2 regularizations. The model was trained for 3500 epochs and it incorporated the utilization of different optimizers like Adam, SGD, RMSprop, Adagrad, Adadelta and utilized a binary cross-entropy loss function. The dataset used includes fourteen features vital in determining the condition of the machine and the risk of a failure. Amongst the different optimizers Adam gave the best results. © 2024 IEEE.

关键词： Cost reduction

来源：评论

学校读者我要写书评

暂无评论

Federated Foundation Models on Heterogeneous Time Series 39

Federated Foundation Models on Heterogeneous Time Series

引用

39th Annual AAAI Conference on artificial intelligence, AAAI 2025

作者： Chen, Shengchao Long, Guodong Jiang, Jing Zhang, Chengqi Australian Artificial Intelligence Institute University of Technology Sydney Australia Department of Data Science and Artificial Intelligence The Hong Kong Polytechnic University Hong Kong

ISBN: (纸本)157735897X

Training a general-purpose time series foundation models with robust generalization capabilities across diverse applications from scratch is still an open challenge. Efforts are primarily focused on fusing cross-domain time series datasets to extract shared subsequences as tokens for training models on Transformer architecture. However, due to significant statistical heterogeneity across domains, this cross-domain fusing approach doesn’t work effectively as the same as fusing texts and images. To tackle this challenge, this paper proposes a novel federated learning approach to address the heterogeneity in time series foundation models training, namely FFTS. Specifically, each data-holding organization is treated as an independent client in a collaborative learning framework with federated settings, and then many client-specific local models will be trained to preserve the unique characteristics per dataset. Moreover, a new regularization mechanism will be applied to both client-side and server-side, thus to align the shared knowledge across heterogeneous datasets from different domains. Extensive experiments on benchmark datasets demonstrate the effectiveness of the proposed federated learning approach. The newly learned time series foundation models achieve superior generalization capabilities on cross-domain time series analysis tasks. Copyright © 2025, Association for the Advancement of artificial intelligence (***). All rights reserved.

关键词： Federated learning

来源：评论

学校读者我要写书评

暂无评论

Quora Question Duplication Detection: An ML Approach for Identifying Semantically Equivalent Questions 1st

Quora Question Duplication Detection: An ML Approach for Ide...

引用

1st International Conference on artificial intelligence Towards Smart Sustainable Society, AISSS 2023

作者： Mundargi, Zarinabegam Kalal, Sahil Kashyap, Aryan Kamble, Omkar Pande, Soham Mhamane, Saniya Department of Artificial Intelligence and Data Science Vishwakarma Institute of Technology Pune411037 India

ISBN: (纸本)9789819775910

This research paper delves into the field of Natural Language Understanding (NLU) by tackling the problem of identifying duplicate questions, using the Quora dataset as the primary source of data. Our study involves a comprehensive examination of the dataset and the application of machine learning techniques, specifically Random Forest and XGBoost models. Importantly, our results emphasize the efficiency of a straightforward CBOW neural network model, outperforming the effectiveness of more complex recurrent and attention-based models. The previous problem on Quora, characterized by numerous duplicate questions, has led to increased ambiguity and a diminished user experience. We also rigorously conduct error analysis, revealing subtleties and subjectivity in the dataset's labeling process. This investigation underscores the value of neural network-based approaches in tackling the complexities of duplicate question detection, which significantly contributes to the broader landscape of NLU research. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Neural network models

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：