检索结果-内蒙古大学图书馆

computer Graphics and Image Processing (CGIP), International Conference on

作者： Trassandra Jewelle Ipapo Charlize Del Rosario Raphael Alampay Patricia Angela Abu Dept. of Information Systems and Computer Science Ateneo Laboratory for Intelligent Visual Environments Ateneo de Manila University Quezon City Philippines

Class imbalance, which negatively affects classification model performance, is a common problem with machine learning. Various oversampling methods have been developed as potential solutions to compensate for imbalanced data. SMOTE is one of the more common methods employed. However, deep generative models such as the variational autoencoder are showing promise as alternatives to traditional oversampling methods. This study investigated the potential of variational autoencoders in learning the distribution of the minority class and producing new observations of facial motion features extracted from an imbalanced medical dataset as well as to see the effects of oversampling before and after the train-test split. The effectiveness of the variational autoencoder was compared to SMOTE in increasing ordinal classification performance across the metrics of accuracy, accuracy±1, inter-rater reliability, specificity, and sensitivity with no oversampling serving as the baseline. The results show that the variational autoencoder has potential as an oversampling method for facial motion features in the context of oro-facial dysfunction estimation. Oversampling prior to the train-test split was also shown to improve classification performance.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Enhancing Communication through Speech to Sign Language Converter

Enhancing Communication through Speech to Sign Language Conv...

引用

North Karnataka Subsection Flagship International Conference (NKCon), IEEE

作者： Gayana M N Mohammed Rumaiz Shaz Rasheed Hussain Shravya A Shreenath Acharya Sana Parveen Dept. of Intelligent Computing & Business Systems St Joseph Engineering College Mangaluru India Dept. of Computer Science and Engineering St Joseph Engineering College Mangaluru India

There are millions of people suffering from speaking and hearing disabilities. Sign language is the only way to communicate for such people to convey their message. The advancement in technology has highly impacted this section of society, hence there is a need to deal with them. This paper describes the Application, "Speech to Sign Language Converter" for communicating with people having hearing and speaking impairments. The primary function of the Application is to convert input sentences into Indian Sign Language (ISL) actions in real time. It takes input in the form of speech or text and produces corresponding sign language using an Avatar which acts based on SiGML code. It can also be used to convert PDF based texts into sign language. In addition, the users can learn Indian Sign Language (ISL) using the tutorials provided within it. The application focuses mainly on converting local languages such as Kannada into ISL.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Fine-Tuned Variational Quantum Classifiers for Cyber Attacks Detection Based on Parameterized Quantum Circuits and Optimizers

Fine-Tuned Variational Quantum Classifiers for Cyber Attacks...

引用

IEEE Annual International computer Software and Applications Conference (COMPSAC)

作者： Md Abdur Rahman Mst. Shapna Akter Emily Miller Bogdan Timofti Hossain Shahriar Mohammad Masum Fan Wu Dept. of Intelligent Systems and Robotics University of West Florida Pensacola FL USA Dept. of Computer Science University of West Florida Pensacola FL USA Center for Cybersecurity University of West Florida Pensacola FL USA Dept. of Applied Data Science San Jose State University San Jose CA USA Dept. of Computer Science Tuskegee University Tuskegee AL USA

ISBN: (数字)9798350376968

ISBN: (纸本)9798350376975

Recent investigations into Quantum Machine Learning (QML) techniques have unveiled methodologies that accelerate training in established machine learning models to provide an alternative for capturing complex patterns. This study focuses on implementing a practical QML Algorithm, Variational Quantum Classification (VQC) for cybersecurity dataset so that detecting anomalies can be improved and faster by reducing number of attributes $\log_{2} M$ while training the model using Qiskit. Also, we study quantum algorithms to understand how it impacts on cyber datasets to detect anomalies in a improved way as it follows logarithms in the dimensionality reduction of quantum states which opens new horizons to quantum big data applications. Most importantly, we aim to also investigate the impact of various parameterized quantum circuits on VQC using quantum data as quantum states encoded by the cyber security dataset, NSL-KDD. In this research, we train VQC with various structures and parameters of quantum circuits as well as optimizers to adjust parameters of quantum circuits (ansatz) to minimize the objective function values so as to improve accuracy of the model in which quantum circuit, EfficientSU2, along with optimizer, COBYLA, outperforms the accuracy than other circuits and optimizers which shows great potential for improving cybersecurity systems. The research could effectively bridge in the gap between theory and implementation based quantum machine learning on cybersecurity systems.

关键词： Training Support vector machines Iris Accuracy Computational modeling Quantum state Quantum circuit

来源：评论

学校读者我要写书评

暂无评论

Mitigating Insecure Outputs in Large Language Models(LLMs): A Practical Educational Module

Mitigating Insecure Outputs in Large Language Models(LLMs): ...

引用

IEEE Annual International computer Software and Applications Conference (COMPSAC)

作者： MD Abdul Barek Md Mostafizur Rahman Shapna Akter A.B.M Kamrul Islam Riad Md Abdur Rahman Hossain Shahriar Akond Rahman Fan Wu Department of Intelligent Systems and Robotics University of West Florida USA Dept. of Cybersecurity and Information Technology University of West Florida USA Center for Cybersecurity University of West Florida USA Computer Science and Software Engineering Auburn University USA Dept. of Computer Science Tuskegee University USA

ISBN: (数字)9798350376968

ISBN: (纸本)9798350376975

Large Language Models (LLMs) have extensive ability to produce promising output. Nowadays, people are increasingly relying on them due to easy accessibility, rapid and outstanding outcomes. However, the use of these results without appropriate scrutiny poses serious security risks, particularly when they are integrated with other software, APIs, or plugins. This is because the LLM outputs are highly dependent on the prompts they receive. Therefore, it is essential to carefully clean these outputs before using them in additional software environments. This paper is designed to teach students about the potential dangers of contaminated LLM output within the context of web development through prelab, hands-on, and postlab experiences. Hands-on lab provides practical guidance on how to handle LLM vulnerabilities to make applications safe with some real-world examples in Python. This approach aims to provide students with a deeper understanding of the precautions necessary to ensure software against the vulnerabilities introduced by LLM output.

关键词： Training Navigation Large language models Computational modeling Software Security Software development management

来源：评论

学校读者我要写书评

暂无评论

Variations on a Hexagon: Iterative Design of Interactive Cyberphysical Tokens and Constraints 24

Variations on a Hexagon: Iterative Design of Interactive Cyb...

引用

18th International Conference on Tangible, Embedded, and Embodied Interaction, TEI 2024

作者： Ullmer, Brygg Dai, Sida De Siqueira, Alexandre Gomes McLendon, Millon Filipiak, Breanna Shafiee, Laila Newman, Winifred Elysse Konkel, Miriam K. Tangible Visualization Group Clemson University United States Virtual Environments Research Group University of Florida United States Computer Science Division School of Computing Clemson University United States Institute for Intelligent Materials Systems and Environments Clemson University United States Dept. of Genetics and Biochemistry Center for Human Genetics Clemson University United States

ISBN: (纸本)9798400704024

We describe iterations on the design of a hexagonal token and constraint tangible and cyberphysical interface. Our system has been co-designed both for use in tangible, embedded, embodied interaction and extended reality classroom contexts, and within broader applied generalizations. We survey some of the related literature in hexagonal, rectangular, and triangular tangible interfaces. We express a design space characterizing several aspects of these prior systems;and apply these to a number of additional virtual and physical iterations. We accompany our text with 3D printable, circuit board, and virtual models for a number of these interactors, usable in both tangible and purely virtual forms, including versions engaging four different embedded computing platforms (Raspberry Pi Zero, Raspberry Pi Pico, Adafruit Circuit Playground, and Blinks). We discuss experiences from use of these in classroom settings, and steps toward broader applications. © 2024 Copyright held by the owner/author(s). Publication rights licensed to ACM.

关键词： 3D printing

来源：评论

学校读者我要写书评

暂无评论

Medical Knowledge-Guided Deep Learning for Mammographic Breast Density Classification

Medical Knowledge-Guided Deep Learning for Mammographic Brea...

引用

Medical Imaging 2023: computer-Aided Diagnosis

作者： Gao, Long Liu, Chang Arefan, Dooman Panigrahy, Ashok Zuley, Margarita L. Wu, Shandong Dept. of Radiology University of Pittsburgh PittsburghPA15213 United States Dept. of Bioengineering University of Pittsburgh PittsburghPA15213 United States Dept. of Biomedical Informatics University of Pittsburgh PittsburghPA15213 United States Intelligent Systems Program University of Pittsburgh PittsburghPA15213 United States College of Computer National University of Defense Technology Changsha410073 China Magee-Womens Hospital University of Pittsburgh Medical Center PittsburghPA15213 United States

ISBN: (纸本)9781510660359

Classification of Breast Imaging Reporting and Data System (BI-RADS) breast density categories generally reflects the amount of dense/fibroglandular tissue in the breast. Studies have consistently shown that breast with higher density has a higher risk of developing breast cancer compared to breast with lower density. In this paper, we propose a novel end-to-end method, namely, Medical Knowledge-guided Deep Learning (MKDL), for breast mammogram density classification. The principle behind MKDL lies in the fact that many breast image density classification tasks are partly or largely based on certain pre-known image features, such as image contrast and brightness. These pre-known features can be computationally represented and then leveraged as prior knowledge to facilitate more effective model learning and thus boost the classification performance. We designed specific knowledge-based transformations for breast density classification and showed that our model outperformed several state-of-the-art models. © 2023 SPIE.

关键词： Classification (of information)

来源：评论

学校读者我要写书评

暂无评论

Authentic Learning Approach for Data Poisoning Vulnerability in LLMs

Authentic Learning Approach for Data Poisoning Vulnerability...

引用

IEEE Annual International computer Software and Applications Conference (COMPSAC)

作者： Mst Shapna Akter Md Abdur Rahman Md Mostafizur Rahman Juanjose Rodriguez-Cardenas Hossain Shahriar Fan Wu Muhammad Rahman Dept. of Intelligent Systems and Robotics University of West Florida Florida USA Dept of Cybersecurity and Information Technology University of West Florida Florida USA Dept. of Information Systems and Security Kennesaw State University Kennesaw USA Center for Cybersecurity University of West Florida Florida USA Dept of Computer Science Tuskegee University Tuskegee USA Dept. of Computer Science and Information Technology Clayton State University Georgia USA

ISBN: (数字)9798350376968

ISBN: (纸本)9798350376975

The primary goal of authentic learning is to provide students with an engaging learning environment that offers hands-on experiences in solving real-world security challenges. Each educational theme consists of prelab activities, lab activities, and hands-on lab activities. By implementing authentic learning, we design and build portable lab for data poisoning vulnerability in LLM models on Google Colab. These hands-on labs can be accessed and practiced in real-time without the need for complex installations and configurations. This allows students to focus on learning concepts and increase their hands-on problem-solving skills.

关键词： Software Real-time systems Data models Internet Security Problem-solving

来源：评论

学校读者我要写书评

暂无评论

COULD DROPPING A FEW CELLS CHANGE THE TAKEAWAYS FROM DIFFERENTIAL EXPRESSION?

arXiv

引用

arXiv 2023年

作者： Shiffman, Miriam Giordano, Ryan Broderick, Tamara MIT Computational & Systems Biology Program United States MIT Dept. of Electrical Engineering & Computer Science United States UC Berkeley Dept. of Statistics United States

Analysis of differential gene expression plays a fundamental role in biology toward illuminating the molecular mechanisms driving a difference between groups (e.g., due to treatment or disease). While any analysis is run on particular cells or samples, the intent is to generalize to future occurrences of the treatment or disease. Implicitly, generalization is justified under the assumption that present and future samples are independent and identically distributed from the same population. Though this assumption is always false, we might hope that any deviation from the assumption is small enough that A) fundamental conclusions of the analysis still hold, and B) standard tools like standard error, significance, and power still reflect generalizability. Conversely, we might worry about these deviations, and reliance on standard statistical tools, if conclusions could be substantively changed by dropping a very small fraction of observations. While checking every small fraction is computationally intractable, recent work develops an approximation to identify when such an influential subset exists. Building on this work, we develop a metric for dropping-data robustness of differential expression;namely, we cast the analysis in a differentiable form suitable to the approximation, extend the approximation to models whose hyperparameters depend on the full dataset, and extend the notion of a data point from a single cell to a pseudobulk observation. We then overcome the inherent non-differentiability of gene set enrichment analysis to develop an additional approximation for the robustness of top gene sets. We use our tool to assess the robustness of differential expression for published single-cell RNA-seq data, and discover that thousands of genes can have their results flipped by dropping Copyright © 2023, The Authors. All rights reserved.

关键词： Diseases

来源：评论

学校读者我要写书评

暂无评论

Developmental Scaffolding with Large Language Models

Developmental Scaffolding with Large Language Models

引用

IEEE International Conference on Development and Learning, ICDL

作者： Batuhan Celik Alper Ahmetoglu Emre Ugur Erhan Oztop Computer Engineering Dept. Bogazici University Istanbul Turkey Computer Science Dept. Ozyegin University Istanbul Turkey Symbiotic Intelligent Systems Research Center Institute for Open and Transdisciplinary Research Initiatives Osaka University Japan

Exploration and self-observation are key mechanisms of infant sensorimotor development. These processes are further guided by parental scaffolding to accelerate skill and knowledge acquisition. In developmental robotics, this approach has been adopted often by having a human acting as the source of scaffolding. In this study, we investigate whether Large Language Models (LLMs) can act as a scaffolding agent for a robotic system that aims to learn to predict the effects of its actions. To this end, an object manipulation setup is considered where one object can be picked and placed on top of or in the vicinity of another object. The adopted LLM is asked to guide the action selection process through algorithmically generated state descriptions and action selection alternatives in natural language. The simulation experiments that include cubes in this setup show that LLM-guided (GPT3.5-guided) learning yields significantly faster discovery of novel structures compared to random exploration. However, we observed that GPT3.5 fails to effectively guide the robot in generating structures with different affordances such as cubes and spheres. Overall, we conclude that even without fine-tuning, LLMs may serve as a moderate scaffolding agent for improving robot learning, however, they still lack affordance understanding which limits the applicability of the current LLMs in robotic scaffolding tasks.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Developmental Scaffolding with Large Language Models

arXiv

引用

arXiv 2023年

作者： Celik, Batuhan Ahmetoglu, Alper Ugur, Emre Oztop, Erhan Computer Engineering Dept. Bogazici University Istanbul Turkey Computer Science Dept. Ozyegin University Istanbul Turkey Symbiotic Intelligent Systems Research Center Institute for Open and Transdisciplinary Research Initiatives Osaka University Japan

关键词： Scaffolds

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：