检索结果-内蒙古大学图书馆

ChatGPT Code Detection: Techniques for Uncovering the Source of Code

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Oedingen, Marc Engelhardt, Raphael C. Denz, Robin Hammer, Maximilian Konen, Wolfgang Cologne Institute of Computer Science Faculty of Computer Science and Engineering Science TH Köln Gummersbach Germany Department of Medical Informatics Biometry and Epidemiology Faculty of Medicine Ruhr-University Bochum Bochum Germany

In recent times, large language models (LLMs) have made significant strides in generating computer code, blurring the lines between code created by humans and code produced by artificial intelligence (AI). As these technologies evolve rapidly, it is crucial to explore how they influence code generation, especially given the risk of misuse in areas like higher education. This paper explores this issue by using advanced classification techniques to differentiate between code written by humans and that generated by ChatGPT, a type of LLM. We employ a new approach that combines powerful embedding features (black-box) with supervised learning algorithms – including Deep Neural Networks, Random Forests, and Extreme Gradient Boosting – to achieve this differentiation with an impressive accuracy of 98%. For the successful combinations, we also examine their model calibration, showing that some of the models are extremely well calibrated. Additionally, we present white-box features and an interpretable Bayes classifier to elucidate critical differences between the code sources, enhancing the explainability and transparency of our approach. Both approaches work well but provide at most 85 − 88% accuracy. We also show that untrained humans solve the same task not better than random guessing. This study is crucial in understanding and mitigating the potential risks associated with using AI in code generation, particularly in the context of higher education, software development, and competitive programming. © 2024, CC BY.

关键词： Deep neural networks

DEPTH-GUIDED FREE-SPACE SEGMENTATION FOR A MOBILE ROBOT

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Sevastopoulos, Christos Hussain, Joey Konstantopoulos, Stasinos Karkaletsis, Vangelis Makedon, Fillia Department of Computer Science & Computer Engineering University of Texas at Arlington ArlingtonTX United States Institute of Informatics and Telecommunications NCSR'Demokritos' Agia Paraskevi Attica Greece

Accurate indoor free-space segmentation is a challenging task due to the complexity and the dynamic nature that indoor environments exhibit. We propose an indoors free-space segmentation method that associates large depth values with navigable regions. Our method leverages an unsupervised masking technique that, using positive instances, generates segmentation labels based on textural homogeneity and depth uniformity. Moreover, we generate superpixels corresponding to areas of higher depth and align them with features extracted from a Dense Prediction Transformer (DPT). Using the estimated free-space masks and the DPT feature representation, a SegFormer model is fine-tuned on our custom-collected indoor dataset. Our experiments demonstrate sufficient performance in intricate scenarios characterized by cluttered obstacles and challenging identification of free space. © 2023, CC BY.

关键词： Mobile robots

Causal Prompting: Debiasing Large Language Model Prompting Based on Front-Door Adjustment 39

学校读者我要写书评

暂无评论

Causal Prompting: Debiasing Large Language Model Prompting B...

39th Annual AAAI Conference on Artificial Intelligence, AAAI 2025

作者： Zhang, Congzhi Zhang, Linhai Wu, Jialong He, Yulan Zhou, Deyu School of Computer Science and Engineering Key Laboratory of Computer Network and Information Integration Ministry of Education Southeast University China Department of Informatics King’s College London United Kingdom The Alan Turing Institute United Kingdom

ISBN: (纸本)157735897X

Despite the notable advancements of existing prompting methods, such as In-Context Learning and Chain-of-Thought for Large Language Models (LLMs), they still face challenges related to various biases. Traditional debiasing methods primarily focus on the model training stage, including approaches based on data augmentation and reweighting, yet they struggle with the complex biases inherent in LLMs. To address such limitations, the causal relationship behind the prompting methods is uncovered using a structural causal model, and a novel causal prompting method based on front-door adjustment is proposed to effectively mitigate LLMs biases. In specific, causal intervention is achieved by designing the prompts without accessing the parameters and logits of LLMs. The chain-of-thought generated by LLM is employed as the mediator variable and the causal effect between input prompts and output answers is calculated through front-door adjustment to mitigate model biases. Moreover, to accurately represent the chain-of-thoughts and estimate the causal effects, contrastive learning is used to fine-tune the encoder of chain-of-thought by aligning its space with that of the LLM. Experimental results show that the proposed causal prompting approach achieves excellent performance across seven natural language processing datasets on both open-source and closed-source LLMs. © 2025, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Contrastive Learning

Cultivate Smart and Healthy Ageing 9th

学校读者我要写书评

暂无评论

Cultivate Smart and Healthy Ageing

9th EAI International Conference on IoT Technologies for HealthCare, HealthyIoT 2022

作者： Andreou, Andreas Mavromoustakis, Constandinos X. Markakis, Evangelos Mastorakis, George Batalla, Jordi Mongay Pallis, Evangelos Dobre, Ciprian Department of Computer Science University of Nicosia and University of Nicosia Research Foundation Nicosia Cyprus Department of Electrical and Computer Engineering Hellenic Mediterranean University Heraklion Crete Greece Department of Management Science and Technology Hellenic Mediterranean University Agios Nikolaos Crete Greece Warsaw University of Technology Warsaw Poland Department of Industrial Design and Production Engineering School of Engineering University of West Attica Athens Greece National Institute for Research and Development in Informatics Bucharest Romania

ISBN: (纸本)9783031286629

The essential criterion for developing an intelligent age-friendly environment is evaluating and monitoring the provision of services. Our research aims to allow the deployment of a broad range of digital healthcare solutions interconnected in an IoT network ecosystem. Through their interoperability, we will be able to cultivate age-friendly smart homes for older adults. Thus, we will enable interpreted living and wellbeing. The intelligent integrations of digital solutions will allow the acquisition and evaluation of health-related data. Standardization, interoperability and scalability of the integration will increase efficiency in healthcare, improving older adults’ Quality of Life (QoL). Also, interpreting their needs will allow key stakeholders to optimize the Quality of Service (QoS). The proposed Internet of Health (IoH) framework will encapsulate sustainable and affordable innovative solutions and their benefits for a comfortable, meaningful, and independent life in an intelligent environment. Therefore, the frame mediated by edge computing, in-home and community settings will be able to interact with healthcare networks contributing to hospitalizations and institutional care. © 2023, ICST institute for computer Sciences, Social informatics and Telecommunications engineering.

关键词： Internet of things

A survey and recent advances in black widow optimization: variants and applications

学校读者我要写书评

暂无评论

Neural Computing and Applications 2024年 1-21页

作者： Shehab, Mohammad Shambour, Mohd Khaled Yousef Abu Hashem, Muhannad A. Al Hamad, Husam Ahmad Shannaq, Fatima Mizher, Manar Jaradat, Ghaith Sh. Daoud, Mohammad Abualigah, Laith College of Computer Sciences and Informatics Amman Arab University Amman11953 Jordan The Custodian of the Two Holy Mosques Institute for Hajj and Umrah Research Umm Al-Qura University Makkah Saudi Arabia Department of Geomatics Architecture and Planning Faculty King Abdulaziz University Jeddah Saudi Arabia College of Engineering Al Ain University Abu Dhabi United Arab Emirates Research Center University of Tabuk Tabuk71491 Saudi Arabia Computer Science Department Al al-Bayt University Mafraq25113 Jordan Hourani Center for Applied Scientific Research Al-Ahliyya Amman University Amman19328 Jordan MEU Research Unit Middle East University Amman11831 Jordan Department of Electrical and Computer Engineering Lebanese American University Byblos13-5053 Lebanon School of Computer Sciences Universiti Sains Malaysia Pulau Pinang11800 Malaysia School of Engineering and Technology Sunway University Malaysia Petaling Jaya27500 Malaysia Applied science research center Applied science private university Amman11931 Jordan College of Engineering Yuan Ze University Taoyuan32003 Taiwan

Artificial Intelligence techniques, such as optimization algorithms, have become essential for success in many fields. Therefore, most researchers, especially in computer and engineering sciences, focused their efforts and abilities on adapting the optimization algorithms for solving various problems. This review introduces one of the recent nature-inspired optimization algorithms, called black widow optimization (BWO). This review highlights the main procedures of BWO, its strengths and weaknesses, the growth of BWO, the variants of BWO (i.e., binary, modification, and hybridization), and the efficiency of BWO in different fields and applications (i.e., feature selection, medical, image processing, engineering, IoT, power and energy, and networks). As a result, the review seeks to assist researchers who are interested in using the BWO to address their challenges. For instance, it could help manage large pilgrim crowds, enabling smoother ritual performances and overcoming potential difficulties they encounter. Also, it suggests new possible ways of enhancing the efficiency of the basic BWO. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Optimization

Audio-Driven Talking Face Generation with Diverse yet Realistic Facial Animations

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Wu, Rongliang Yu, Yingchen Zhan, Fangneng Zhang, Jiahui Zhang, Xiaoqin Lu, Shijian School of Computer Science and Engineering Nanyang Technological University Singapore Max Planck Institute for Informatics Germany College of Computer Science and Artifcial Intelligence Wenzhou University China

Audio-driven talking face generation, which aims to synthesize talking faces with realistic facial animations (including accurate lip movements, vivid facial expression details and natural head poses) corresponding to the audio, has achieved rapid progress in recent years. However, most existing work focuses on generating lip movements only without handling the closely correlated facial expressions, which degrades the realism of the generated faces greatly. This paper presents DIRFA, a novel method that can generate talking faces with diverse yet realistic facial animations from the same driving audio. To accommodate fair variation of plausible facial animations for the same audio, we design a transformer-based probabilistic mapping network that can model the variational facial animation distribution conditioned upon the input audio and autoregressively convert the audio signals into a facial animation sequence. In addition, we introduce a temporally-biased mask into the mapping network, which allows to model the temporal dependency of facial animations and produce temporally smooth facial animation sequence. With the generated facial animation sequence and a source image, photo-realistic talking faces can be synthesized with a generic generation network. Extensive experiments show that DIRFA can generate talking faces with realistic facial animations effectively. Copyright © 2023, The Authors. All rights reserved.

关键词： Mapping

EfficientNetB0-based Automated Diabetic Retinopathy Classification in Fundus Images

学校读者我要写书评

暂无评论

EfficientNetB0-based Automated Diabetic Retinopathy Classifi...

Automation, Computing and Renewable Systems (ICACRS), International Conference on

作者： Ezekiel Alaric Majaw G. Naveen Sundar D. Narmadha Senthil Kumar Thangavel Venkatesan Adeyemi Abel Ajibesin Department of Computer Science and Engineering Karunya Institute of Technology and Sciences Coimbatore India Department of Computer Science and Engineering Amrita School of Computing Coimbatore India Department of Information Technology Faculty of Informatics and Design Cape Peninsula University of Technology South Africa

ISBN: (数字)9798331532420

ISBN: (纸本)9798331532437

The most common cause of blindness among diabetics is diabetic retinopathy. This disease must be detected in its early stages as delaying treatment can result in permanent blindness. In today’s modern world with the development of technologies, automated techniques have improved the accuracy and efficiency of detecting and classifying this disease. This study suggests a deep learning model utilizing EfficientNetB0 that focuses on designing a neural network structure that requires fewer parameters and computations compared to traditional architectures. Further to help in increasing the accuracy, the image is being processed with methods like CLAHE which is an image processing algorithm that helps in enhancing the contrast while keeping the noise of the images as low as possible. Also, augmentation is being done on the images and SMOTE is applied to handle class imbalance. We collected the APTOS Blindness Detection dataset consisting of retinal fundus images. The accuracy that is being obtained from the proposed methodology is 98.78% indicating the model's effectiveness in detection and classifying between the various stages of diabetic retinopathy.

关键词： Diabetic retinopathy Renewable energy sources Accuracy Computational modeling Blindness computer architecture Transformers Retina Real-time systems Diseases

Randomness-Efficient Constructions of Capacity-Achieving List-Decodable Codes

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Mosheiff, Jonathan Resch, Nicolas Shang, Kuo Yuan, Chen Department of Computer Science Ben-Gurion University Informatics Institute University of Amsterdam Netherlands School of Electronic Information and Electrical Engineering Shanghai Jiao Tong University China

In this work, we consider the task of generating list-decodable codes over small (say, binary) alphabets using as little randomness as possible. Specifically, we hope to generate codes achieving what we term the Elias bound, which means that they are (ρ, L)-list-decodable with rate R ≥ 1 − h(ρ) − O(1/L). A long line of work shows that uniformly random linear codes (RLCs) achieve the Elias bound: hence, we know O(n2) random bits suffice. Prior works (Guruswami and Mosheiff, FOCS 2022;Putterman and Pyne, arXiv 2023) demonstrate that just O(Ln) random bits suffice, via puncturing of low-bias codes. These recent constructions are essentially combinatorial, and rely (directly or indirectly) on graph expansion. We provide two new constructions, which are algebraic. Compared to prior works, our constructions are considerably simpler and more direct. Furthermore, our codes are designed in such a way that their duals are also quite easy to analyze. Our first construction — which can be seen as a generalization of the celebrated Wozencraft ensemble — achieves the Elias bound and consumes Ln random bits. Additionally, its dual code achieves the Gilbert-Varshamov bound with high probability, and both the primal and dual admit quasilinear-time encoding algorithms. The second construction consumes 2nL random bits and yields a code where both it and its dual achieve the Elias bound. As we discuss, properties of a dual code are often crucial for applications of error-correcting codes in cryptography. In all of the above cases – including the prior works achieving randomness complexity O(Ln) – the codes are designed to "approximate" RLCs. More precisely, for a given locality parameter L we construct codes achieving the same L-local properties as RLCs. This allows one to appeal to known list-decodability results for RLCs and thereby conclude that the code approximating an RLC also achieves the Elias bound (with high probability). As a final contribution, we indicate that such a proof

关键词： Random processes

Distributed Secret Securing in Discrete-Event Systems

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Matsui, Shoma Cai, Kai Rudie, Karen Department of Electrical and Computer Engineering Ingenuity Labs Research Institute Queen’s University Kingston Canada Department of Core Informatics Osaka Metropolitan University Osaka Japan

In this paper, we study a security problem of protecting secrets in distributed systems. Specifically, we employ discrete-event systems to describe the structure and behaviour of distributed systems, in which global secret information is separated into pieces and stored in local component agents. The goal is to prevent such secrets from being exposed to intruders by imposing appropriate protection measures. This problem is formulated as to ensure that at least one piece of every distributed global secret is secured by a required number of protections, while the overall cost to apply protections is minimum. We first characterize the solvability of this security problem by providing a necessary and sufficient condition, and then develop an algorithm to compute a solution based on the supervisory control theory of discrete-event systems. Finally, we illustrate the effectiveness of our solution with an example system comprising distributed databases. Copyright © 2024, The Authors. All rights reserved.

关键词： Network security