检索结果-内蒙古大学图书馆

International Conference on ICT For Smart Society (ICISS)

作者： Felicia Ferren Felix Indra Kurniadi Statistics Department School of Computer Science Bina Nusantara University Jakarta Indonesia Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia

Since the 1960s, bankruptcy prediction research has been intensively examined, with the importance of accurate forecasting representing a nation's economic development. The high costs related to business failures have prompted efforts to enhance prediction techniques, which have been highlighted further as the effects of globalization have increased. There are two primary types of bankruptcy prediction models: statistical and artificial intelligence (AI) methods. The study found that several AI algorithms, namely C5.0, CART, and Support Vector Machines with evolutionary computation, demonstrated efficacy in forecasting impending bankruptcies, as well as short- and long-term predictions. Prior studies have proposed the utilization of Multilayer Perceptron (MLP), ensemble learning methodologies, and AdaBoost as potential approaches for the purpose of bankruptcy prediction. The significance of feature importance is frequently disregarded in such models, resulting in prejudiced outcomes. This study presents a proposed approach for feature selection in AdaBoost algorithm that incorporates the consideration of feature importance, with the aim of enhancing the aforementioned findings. The proposed methodology is validated through a comparative analysis with several other machine learning algorithms. The result showed that ADABoost with feature selection using feature importance generated better value with increased around 5-7% in accuracy score.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Feature Selection using Grey Wolf Optimization Algorithm on Light Gradient Boosting Machine

Feature Selection using Grey Wolf Optimization Algorithm on ...

引用

computer science, Information Technology and Engineering (ICCoSITE), International Conference on

作者： Felix Indra Kumiadi Ajeng Wulandari Samsul Arifin Computer Science Department School of Computer Science Bina Nusantara University Jakarta Indonesia Statistics Department School of Computer Science Bina Nusantara University Jakarta Indonesia

high dimensional data provide a major problem to supervised learning. In identifying high dimensional data, the learning models usually exhibit overfitting and become less understandable. One way to find the ideal features on high-dimensional data implemented feature selection on dataset Feature selection is one of the crucial aspects on data preprocessing step. Several algorithms for feature selection were proposed over the decades such as wrapper method, filter, and embedded method. In this research, we implemented wrapper method with Grey Wolf Optimization. We implemented Grey Wolf Optimization on wrapper method because the algorithm is efficient, simple and had lower computational time. We are also compared Grey Wolf Optimization to other meta-heuristic algorithms such as Particle Swarm Optimization and Genetic Algorithms. The result showed the GWO provide better computational time with the average time from four different dataset was 6.1125s. The accuracy result showed the GWO performed better on Ionosphere dataset.

关键词： Metaheuristics data preprocessing Filtering algorithms Feature extraction Boosting data models Particle swarm optimization

来源：评论

学校读者我要写书评

暂无评论

Two-sided Competing Matching Recommendation Markets With Quota and Complementary Preferences Constraints

arXiv

引用

arXiv 2023年

作者： Li, Yuantong Cheng, Guang Dai, Xiaowu Department of Statistics and Data Science UCLA United States Department of Biostatistics UCLA United States

In this paper, we propose a new recommendation algorithm for addressing the problem of two-sided online matching markets with complementary preferences and quota constraints, where agents’ preferences are unknown a priori and must be learned from data. The presence of mixed quota and complementary preferences constraints can lead to instability in the matching process, making this problem challenging to solve. To overcome this challenge, we formulate the problem as a bandit learning framework and propose the Multi-agent Multi-type Thompson Sampling (MMTS) algorithm. The algorithm combines the strengths of Thompson Sampling for exploration with a new double matching technique to provide a stable matching outcome. Our theoretical analysis demonstrates the effectiveness of MMTS as it can achieve stability and has a total O(Q√KmaxT)-Bayesian regret with high probability, which exhibits linearity with respect to the total firm’s quota Q, the square root of the maximum size of available type workers √Kmax and time horizon T. In addition, simulation studies also demonstrate MMTS’ effectiveness in various settings. We provide code used in our experiments https://***/Likelyt/Double-Matching. Copyright © 2023, The Authors. All rights reserved.

关键词： Multi agent systems

来源：评论

学校读者我要写书评

暂无评论

Memory-based optimization methods for model-agnostic meta-learning and personalized federated learning

引用

The Journal of Machine Learning Research 2023年第1期24卷 6926-6971页

作者： Bokun Wang Zhuoning Yuan Yiming Ying Tianbao Yang Department of Computer Science and Engineering Texas A&M University College Station TX Department of Computer Science The University of Iowa Iowa City IA Department of Mathematics and Statistics University at Albany Albany NY

In recent years, model-agnostic meta-learning (MAML) has become a popular research area. However, the stochastic optimization of MAML is still underdeveloped. Existing MAML algorithms rely on the "episode" idea by sampling a few tasks and data points to update the meta-model at each iteration. Nonetheless, these algorithms either fail to guarantee convergence with a constant mini-batch size or require processing a large number of tasks at every iteration, which is unsuitable for continual learning or cross-device federated learning where only a small number of tasks are available per iteration or per round. To address these issues, this paper proposes memory-based stochastic algorithms for MAML that converge with vanishing error. The proposed algorithms require sampling a constant number of tasks and data samples per iteration, making them suitable for the continual learning scenario. Moreover, we introduce a communication-efficient memory-based MAML algorithm for personalized federated learning in cross-device (with client sampling) and cross-silo (without client sampling) settings. Our theoretical analysis improves the optimization theory for MAML, and our empirical results corroborate our theoretical findings. Interested readers can access our code at https://***/bokun-wang/moml.

关键词： meta-learning federated learning model-agnostic meta-learning personalized federated learning memory-based algorithms

来源：评论

学校读者我要写书评

暂无评论

Enhancing UAV Path Planning Efficiency through Adam-Optimized Deep Neural Networks for Area Coverage Missions

引用

Procedia computer science 2024年 235卷 2-11页

作者： Akshya J Neelamegam G C. Sureshkumar Nithya V Seifedine Kadry SRM Institute of Science and Technology Department of Computational Intelligence Chennai 603203 India Karpagam College of Engineering Department of Computer Science and Technology Coimbatore 641032 India KGiSL Institute of Technology Department of Information Technology Coimbatore 641032 India KGiSL Institute of Technology Department of Computer Science and Engineering Coimbatore 641032 India Noroff University College Department of Applied Data Science Kristiansand 4612 Norway

Efficient Unmanned Aerial Vehicle (UAV) trajectory generation is crucial for successful area coverage missions, aiming to maximize coverage while minimizing resource consumption. In this research, we present a comprehensive study on optimizing UAV trajectory generation using Deep Neural Networks (DNNs) with the Adam optimization algorithm. The DNNs are trained on historical data to produce smooth and continuous trajectories, thereby reducing abrupt changes in direction and enhancing overall efficiency during the mission. To evaluate the performance of the proposed approach, we conducted experiments comparing different activation functions, namely tanh, sigmoid, and ReLU, with the Adam-optimized DNN model. The trajectories generated by each activation function were analyzed using key metrics, including Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R-squared (R 2 ) scores for both X and Y coordinates. The results of the comparative analysis revealed that the DNN model with the Adam optimizer exhibited superior performance over the other activation functions. It achieved lower MSE, MAE, and RMSE values, indicating better trajectory accuracy and smoother paths. Additionally, the R 2 scores demonstrated a higher correlation between the generated trajectories and the actual trajectories, highlighting the model's ability to capture underlying patterns effectively. The findings underscore the significance of leveraging the Adam-optimized DNN approach for UAV trajectory planning, offering promising opportunities for resource optimization, increased mission success, and further advancements in autonomous aerial systems. This research contributes to the ongoing efforts in UAV path planning, optimization, and intelligent control strategies, paving the way for enhanced autonomous systems in various real-world applications.

关键词： Unmanned Aerial Vehicles (UAVs) Trajectory Generation Deep Neural Networks (DNNs) Adam Optimization Area Coverage Missions

来源：评论

学校读者我要写书评

暂无评论

State-space models with layer-wise nonlinearity are universal approximators with exponential decaying memory 23

State-space models with layer-wise nonlinearity are universa...

引用

Proceedings of the 37th International Conference on Neural Information Processing Systems

作者： Shida Wang Beichen Xue Department of Mathematics National University of Singapore Department of Statistics and Data Science National University of Singapore

State-space models have gained popularity in sequence modelling due to their simple and efficient network structures. However, the absence of nonlinear activation along the temporal direction limits the model's capacity. In this paper, we prove that stacking state-space models with layer-wise nonlinear activation is sufficient to approximate any continuous sequence-to-sequence relationship. Our findings demonstrate that the addition of layer-wise nonlinear activation enhances the model's capacity to learn complex sequence patterns. Meanwhile, it can be seen both theoretically and empirically that the state-space models do not fundamentally resolve the issue of exponential decaying memory. Theoretical results are justified by numerical verifications.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Low-Rank Expectile Representations of a data Matrix, with Application to Diurnal Heart Rates

arXiv

引用

arXiv 2024年

作者： Ouyang, Shuge Tang, Yunxuan Agyare, Benjamin Osafo Department of Statistics University of Michigan United States Department of Computer Science Engineering University of Michigan United States

Low-rank matrix factorization is a powerful tool for understanding the structure of 2-way data, and is usually accomplished by minimizing a sum of squares criterion. Expectile analysis generalizes squared-error loss by introducing asymmetry, allowing tail behavior to be elicited. Here we present a framework for low-rank expectile analysis of a data matrix that incorporates both additive and multiplicative effects, utilizing expectile loss, and accommodating arbitrary patterns of missing data. The representation can be fit with gradient-descent. Simulation studies demonstrate the accuracy of the structure recovery. Using diurnal heart rate data indexed by person-days versus minutes within a day, we find divergent behavior for lower versus upper expectiles, with the lower expectiles being much more stable within subjects across days, while the upper expectiles are much more variable, even within subjects. Copyright © 2024, The Authors. All rights reserved.

关键词： Matrix factorization

来源：评论

学校读者我要写书评

暂无评论

Analyzing and contrasting machine learning algorithms for Intrusion Detection System 24

Analyzing and contrasting machine learning algorithms for In...

引用

24th International Arab Conference on Information Technology, ACIT 2023

作者： Al-Daja, Shrouq Ala'alyabrodi Al-Mousa, Mohammad Rasmi Al-Qammaz, Abdullah Olimat, Kholod Naser Olemat, Hamza M Daoud, Mohammad Sh Zarqa University Department of Data Science and Artificial Intelligence Zarqa Jordan Zarqa University Department of Cyber Security Zarqa Jordan Zarqa University Department of English and Literature Zarqa Jordan Zarqa University Department of Computer Science Zarqa Jordan College of Engineering Al Ain University Abu Dhabi United Arab Emirates

ISBN: (纸本)9798350384307

As the prevalence of sophisticated network attacks continues to rise, enhancing conventional intrusion detection systems (IDS) methods presents a significant challenge. Machine learning (ML)-based anomaly detection and cybersecurity analytics have been widely used in research and development projects related to network IDS. This study explores a set of classifiers on specific domain is a cybersecurity. A significant aspect of this study involves data preparation, including a selection of important features to enhance detection accuracy. Additionally, the study introduces the Synthetic Minority Over-sampling Technique (SMOTE) to address the challenge of classifying imbalanced data. The primary aim of this study is to evaluate various classifiers, such as Random Forest (RF), Decision Tree (DT), K-Nearest Neighbors (KNN), and Multi-layer perceptron (MLP) neural network. The experiments involve different feature groups distributed at percentages of 25%, 50%, 75%, and 100%. The evaluation is conducted from the perspectives of evaluation measurements, time consumption, and space utilization. The experimental findings highlight that RF outperforms the other classifiers, achieving an accuracy of 0.96778, precision of 0.73007, recall of 0.74087, F1 score of 0.73184, and G-mean of 0.22136. © 2023 IEEE.

关键词： CICIDS 2017 cyberattack intrusion detection system machine learning random forest SMOTE

来源：评论

学校读者我要写书评

暂无评论

Fostering Computational Thinking in Elementary Mathematics Instruction and Learning with the Support of Large Language Models 2025

Fostering Computational Thinking in Elementary Mathematics I...

引用

Proceedings of the 30th ACM Conference on Innovation and Technology in computer science Education V. 2

作者： Emma S. Ylagan Heena P. Parekh Malisha Das Mai Dahshan Department of Computer Science University of Virginia Charlottesville Virginia USA School of Data Science University of Virginia Charlottesville Virginia USA

ISBN: (纸本)9798400715693

Computational thinking (CT) integration in elementary mathematics engages young learners in the decomposition of complex problems and the construction of iterative approaches to mathematical thinking. To effectively integrate MATH+CT, we need to build teachers' capacity in developing knowledge of CT concepts and creating Math+CT activities. This can positively influence students' mathematical outcomes and their readiness for computer science (CS) in middle and high school. Although various professional development programs aim to build teachers' CT knowledge, limited research exists on how teachers apply this knowledge to classroom-based Math+CT activities. Simultaneously, the rapid improvement of large Language Models (LLMs) creates a catalyst for building an innovative resource space to support elementary teachers' integration of MATH+CT in their existing school or district curriculum. In this poster, we present a tool that leverages LLMs to support teachers in creating Scratch programs designed to explore and deepen students' understanding of mathematical concepts.

关键词： computational thinking

来源：评论

学校读者我要写书评

暂无评论

Knowledge Enhancement using Question Generation for Images and Chart data Input 4

Knowledge Enhancement using Question Generation for Images a...

引用

4th International Conference on Artificial Intelligence and Speech Technology, AIST 2022

作者： Tayal, Devendra Kumar Jain, Amita Shrivastava, Nikita Jain, Akshita Gaur, Hunny Indira Gandhi Delhi Technical University for Women Department of Computer Science & Engineering Delhi India Netaji Subhash University of Technology Department of Computer Science & Engineering Delhi India Indira Gandhi Delhi Technical University for Women Department of Artificial Intelligence & Data Science Delhi India University of Delhi St. Stephen's College Department of Computer Science Delhi India

ISBN: (纸本)9781665499026

In the current era, pictorial representation of data in the form of patterns, charts, graphs and trends are gaining a lot of attention in the fields of finance and science. Deep and correct study of the trends going on is necessary to better one's financial and general knowledge and helps reading and comparing the patterns over a period of time. Correct understanding of the pictorial representation of data in the form of line charts, bar charts or any textual representation is a major task. Hence, the proposed method aims at reading these charts and images, converting them into text and generating questions and then using sentiment analysis to determine positive or negative trends. This is followed by generating commonly asked questions based on the extracted text and providing more context and information on this text using web crawling, which helps provide greater dimensionality and understanding of the problem at hand. © 2022 IEEE.

关键词： Sentiment analysis

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：