检索结果-内蒙古大学图书馆

Harnessing Deep Learning for Effective Extractive Text Summarization: A Comparative Study

学校读者我要写书评

暂无评论

Harnessing Deep Learning for Effective Extractive Text Summa...

Intelligent Methods, Systems, and Applications (IMSA)

作者： Deema mohammed Alsekait Waref Almanaseer Ibrahim A Gomaa Magdy Abd-Elghany Zeid Salma Baligh Hana Alaa Jasmine Hegazy Julia Magdy Alaa Tariq Eldmrat Diaa Salama AbdElminaam Department of Computer Science and Information Technology Applied College Princess Nourah Bint Abdulrahman University Saudi Arabia Faculty of Information Technology Applied Science Private University Amman Jordan Computer Science Department Alobour high institute for computer and informatics Cairo Egypt Faculty of Computer Science Misr International University Cairo Egypt Faculty of Human Sciences Midocean University Jadara Research Center Jadara University Irbid Jordan Faculty of Computers Misr International University Egypt. MEU Research Unit Middle East University Amman Jordan

ISBN: (数字)9798350362633

ISBN: (纸本)9798350362640

Text Summarization might seem like a trivial problem, but in reality, is essential for data acquisition and understanding by compressing large amounts of text into specific a nd to-the-point summaries. This research is conducted to help compare the proposed methods' applications and make a summary system accessible to applications such as information retrieval, legal documents summarization and much more. from this, we intend to start by acquiring the dataset, preprocessing and training, and testing the models. In this paper will explore the methods that can be applied on summarization. These methods are Machine learning models like the random forest, naive Bayes, SVM, logistic regression, Deep learning like a neural networks, Transformers-based models like T5, and performance metrics like ROUGE. By trying all these various methods and models on a news article dataset we are going to compare the results from each of them and get the best performance. Finally, the results would be a comparison between all outputs and found that Natural Language Processor. In conclusion, Text Summarization is best done using NLP, and challenges in this project would be due to the diversity of text and context.

关键词： Support vector machines Deep learning Training Semantics Text summarization Transformers Information retrieval

GKT: A Novel Guidance-Based Knowledge Transfer Framework For Efficient Cloud-edge Collaboration LLM Deployment

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yao, Yao Li, Zuchao Zhao, Hai Department of Computer Science and Engineering Shanghai Jiao Tong University China Shanghai Key Laboratory of Trusted Data Circulation and Governance in Web3 China Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Shanghai Jiao Tong University China National Engineering Research Center for Multimedia Software School of Computer Science Wuhan University Wuhan430072 China

The burgeoning size of Large Language Models (LLMs) has led to enhanced capabilities in generating responses, albeit at the expense of increased inference times and elevated resource demands. Existing methods of acceleration, predominantly hinged on knowledge distillation, generally necessitate fine-tuning of considerably large models, posing a challenge for average users. Furthermore, present techniques for expediting inference and reducing costs operate independently. To address these issues, we introduce a novel and intuitive Guidance-based Knowledge Transfer (GKT) framework. This approach leverages a larger LLM as a "teacher" to create guidance prompts, paired with a smaller "student" model to finalize responses. Remarkably, GKT requires no fine-tuning and doesn’t necessitate the teacher and student models to have the same vocabulary, allowing for extensive batch generation to accelerate the process while ensuring user customization. GKT can be seamlessly integrated into cloud-edge collaboration architectures, and is versatile enough for plug-and-play application across various models. When utilizing ChatGPT as teacher model and Llama2-70B as the student model, we can achieve 95.00% of ChatGPT’s performance at 52% of the cost. The results highlight substantial enhancements in accuracy and processing speed on the GSM8K and CSQA datasets, surpassing the performance of using either the student or teacher models in isolation. Our code is publicly available at https://***/Zoeyyao27/GKT © 2024, CC BY.

关键词： Distillation

Quantifying Kinematic Tremor in an NGLY1-Deficient Individual: A Case Study

学校读者我要写书评

暂无评论

Case Reports in Clinical Medicine 2024年第1期13卷 25-36页

作者： Brock Futrell Christopher Alexander Malaya Dacia Martinez Diaz Christian Alfaro Hannah Elizabeth Gustafson Subhalakshmi Chandrasekaran Rhea Mohan Phatak Bernhard Suter Charles Shannon Layne Department of Health and Human Performance University of Houston Houston USA Center for Neuromotor and Biomechanics Research University of Houston Houston USA Department of Computer Science and Engineering Texas A&M University College Station USA Blue Bird Circle Rett Center Texas Children’s Hospital Houston USA Baylor College of Medicine Houston USA Center for NeuroEngineering and Cognitive Science University of Houston Houston USA

NGLY1 Deficiency is an autosomal recessive congenital disorder that has been identified in less than 100 individuals. Most individuals with NGLY1 Deficiency display hyperkinetic movement disorders, including choreiform, athetoid, dystonic myoclonic, dyskinetic, and dysmetric movements. Developing a consistent and concise consensus on the classification and evaluation of tremors is essential to forward the research and treatment of tremors. It has also been reported that some individuals with NGLY1 Deficiency demonstrate tremor, but such tremor has never been formally investigated. The primary objective of this study is to determine if an individual with NGLY1 Deficiency demonstrates an identifiable tremor during a series of arm movements and, if so, describe the frequency and power characteristics of that tremor. Arm movement kinematics were obtained using a 16-camera Vicon system, and time series trajectory waveforms for three planes of a marker placed on the hand were developed. Custom MATLAB scripts were utilized to compute Fast Fourier Transformations of the data within the identified waveform segments. A mean frequency of 2.30 Hz (SD = 1.05) with a mean power of 5.02 |P1(f)| (SD = 4.63) suggests that our participant’s kinematic data did display a persistent tremor in both hands across all tasks and movement planes. Analyses of the reaching hand and the non-reaching hand suggest the participant displayed an action tremor in both postural and intention (kinetic) tremors. Future directions should include assessing additional individuals with NGLY1 Deficiency to determine if the tremor is a distinguishable disorder behavior. Additionally, evaluating other anatomical sites, such as the elbow, head, and lower limbs, would provide further insights into the characteristics of this tremor.

关键词： Tremor NGLY Fast Fourier Transformation Reaching Hyperkinetic Movement

Design lessons from AI’s two grand goals: human emulation and useful applications

学校读者我要写书评

暂无评论

IEEE Transactions on Technology and Society

IEEE Transactions on Technology and Society 2020年第2期1卷 73-82页

作者： Shneiderman, Ben Department of Computer Science University of Maryland at College Park College ParkMD20742 United States Human–Computer Interaction Laboratory University of Maryland at College Park College ParkMD20742 United States

Researchers’ goals shape the questions they raise, collaborators they choose, methods they use, and outcomes of their work. This article offers a fresh vision of artificial intelligence (AI) research by suggesting a simplification to two goals: 1) emulation to understand human abilities to build systems that perform tasks as well as or better than humans and 2) application of AI methods to build widely used products and services. Researchers and developers for each goal can fruitfully work along their desired paths, but this article is intended to limit the problems that arise when assumptions from one goal are used to drive work on the other goal. For example, autonomous humanoid robots are prominent with emulation researchers, but application developers avoid them, in favor of tool-like appliances or teleoperated devices for widely used commercial products and services. This article covers four such mismatches in goals that affect AI-guided application development: 1) intelligent agent or powerful tool;2) simulated teammate or teleoperated device;3) autonomous system or supervisory control;and 4) humanoid robot or mechanoid appliance. This article clarifies these mismatches to facilitate the discovery of workable compromise designs that will accelerate human-centered AI applications research. A greater emphasis on human-centered AI could reduce AI’s existential threats and increase benefits for users and society, such as in business, education, healthcare, environmental preservation, and community safety. © 2020 IEEE. All right reserved.

关键词： Anthropomorphic robots

CGMatch: A Different Perspective of Semi-supervised Learning

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Cheng, Bo Lu, Jueqing Tian, Yuan Zhao, Haifeng Chang, Yi Du, Lan School of Artificial Intelligence Jilin University China International Center of Future Science Jilin University China Engineering Research Center of Knowledge-Driven Human-Machine Intelligence MOE China Faculty of Information Technology Monash University Australia Department of Computer Science Jinling Institute of Technology China

Semi-supervised learning (SSL) has garnered significant attention due to its ability to leverage limited labeled data and a large amount of unlabeled data to improve model generalization performance. Recent approaches achieve impressive successes by combining ideas from both consistency regularization and pseudo-labeling. However, these methods tend to underperform in the more realistic situations with relatively scarce labeled data. We argue that this issue arises because existing methods rely solely on the model’s confidence, making them challenging to accurately assess the model’s state and identify unlabeled examples contributing to the training phase when supervision information is limited, especially during the early stages of model training. In this paper, we propose a novel SSL model called CGMatch, which, for the first time, incorporates a new metric known as Count-Gap (CG). We demonstrate that CG is effective in discovering unlabeled examples beneficial for model training. Along with confidence, a commonly used metric in SSL, we propose a fine-grained dynamic selection (FDS) strategy. This strategy dynamically divides the unlabeled dataset into three subsets with different characteristics: easy-to-learn set, ambiguous set, and hard-to-learn set. By selective filtering subsets, and applying corresponding regularization with selected subsets, we mitigate the negative impact of incorrect pseudo-labels on model optimization and generalization. Extensive experimental results on several common SSL benchmarks indicate the effectiveness of CGMatch especially when the labeled data are particularly limited. Source code is available at https://***/BoCheng-96/CGMatch. Copyright © 2025, The Authors. All rights reserved.

关键词： Self-supervised learning

Enhanced Fine-Tuning of Lightweight Domain-Specific Q&A Model Based on Large Language Models

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhang, Shenglin Zhu, Pengtian Ma, Minghua Wang, Jiagang Sun, Yongqian Li, Dongwen Wang, Jingyu Guo, Qianying Hua, Xiaolei Zhu, Lin Pei, Dan Nankai University China Microsoft United States Tsinghua University China China Mobile Research Institute China Haihe Laboratory of Information Technology Application Innovation China Tianjin Key Laboratory of Software Experience and Human Computer Interaction China Beijing National Research Center for Information Science and Technology China

Large language models (LLMs) excel at general question-answering (Q&A) but often fall short in specialized domains due to a lack of domain-specific knowledge. Commercial companies face the dual challenges of privacy protection and resource constraints when involving LLMs for fine-tuning. This paper propose a novel framework, Self-Evolution, designed to address these issues by leveraging lightweight open-source LLMs through multiple iterative fine-tuning rounds. To enhance the efficiency of iterative fine-tuning, Self-Evolution employ a strategy that filters and reinforces the knowledge with higher value during the iterative process. We employed Self-Evolution on Qwen1.5-7B-Chat using 4,000 documents containing rich domain knowledge from China Mobile, achieving a performance score 174% higher on domain-specific question-answering evaluations than Qwen1.5-7B-Chat and even 22% higher than Qwen1.5-72B-Chat. Self-Evolution has been deployed in China Mobile's daily operation and maintenance for 117 days, and it improves the efficiency of locating alarms, fixing problems, and finding related reports, with an average efficiency improvement of over 18.6%. In addition, we release Self-Evolution framework code in https://***/Zero-Pointer/Self-Evolution. Copyright © 2024, The Authors. All rights reserved.

关键词： Data mining

CapsEnhancer: An Effective Computational Framework for Identifying Enhancers Based on Chaos Game Representation and Capsule Network

学校读者我要写书评

暂无评论

Journal of Chemical Information and Modeling 2024年第14期64卷 5725-5736页

作者： Yao, Lantian Xie, Peilin Guan, Jiahui Chung, Chia-Ru Huang, Yixian Pang, Yuxuan Wu, Huacong Chiang, Ying-Chih Lee, Tzong-Yi Kobilka Institute of Innovative Drug Discovery School of Medicine The Chinese University of HongKong Shenzhen518172 China School of Science and Engineering The Chinese University of HongKong Shenzhen518172 China School of Medicine The Chinese University of HongKong Shenzhen518172 China Department of Computer Science and Information Engineering National Central University Taoyuan320317 Taiwan Division of Health Medical Intelligence Human Genome Center The University of Tokyo Tokyo108-8639 Japan Institute of Bioinformatics and Systems Biology National Yang Ming Chiao Tung University Hsinchu300093 Taiwan National Yang Ming Chiao Tung University Hsinchu300093 Taiwan

Enhancers are a class of noncoding DNA, serving as crucial regulatory elements in governing gene expression by binding to transcription factors. The identification of enhancers holds paramount importance in the field of biology. However, traditional experimental methods for enhancer identification demand substantial human and material resources. Consequently, there is a growing interest in employing computational methods for enhancer prediction. In this study, we propose a two-stage framework based on deep learning, termed CapsEnhancer, for the identification of enhancers and their strengths. CapsEnhancer utilizes chaos game representation to encode DNA sequences into unique images and employs a capsule network to extract local and global features from sequence "images". Experimental results demonstrate that CapsEnhancer achieves state-of-the-art performance in both stages. In the first and second stages, the accuracy surpasses the previous best methods by 8 and 3.5%, reaching accuracies of 94.5 and 95%, respectively. Notably, this study represents the pioneering application of computer vision methods to enhancer identification tasks. Our work not only contributes novel insights to enhancer identification but also provides a fresh perspective for other biological sequence analysis tasks. © 2024 The Authors. Published by American Chemical Society.

关键词： DNA sequences

Development of a compact mouse retinal imaging system using spectral domain optical coherence tomography (SD-OCT) with a full field stimulation channel for recording fast optoretinograms (ORGs) 34

学校读者我要写书评

暂无评论

Development of a compact mouse retinal imaging system using ...

Ophthalmic Technologies XXXIV 2024

作者： Jo, Hang Chan Pijewska, Ewelina Meleppat, Ratheesh K. Jonnal, Ravi S. Kim, Dae Yu Zawadzki, Robert J. Department of Electrical and Computer Engineering Inha University 100 Inha-ro Michuhol-gu Incheon22212 Korea Republic of Dept. of Ophthalmology and Vision Science UC Davis Eye Center SacramentoCA95817 United States UC Davis Eye-Pod Small Animal Ocular Imaging Laboratory Department of Cell Biology and Human Anatomy University of California Davis 4320 Tupper Hall DavisCA95616 United States

ISBN: (纸本)9781510669079

While the a-wave of mouse electroretinogram (ERG) occurs within 50 milliseconds after exposure to light, the optoretinogram (ORG) slower than a 20Hz sampling rate could face limitations in observing immediate morphological changes from the retina. In this study, we develop a compact custom-built mouse ORG system based on spectral domain optical coherence tomography (SD-OCT) for 100Hz~1KHz B-scan rates comprised of 100 kHz A-scans. All the optics of the developed ORG system are designed on a 24 x 24 inches optical breadboard to move easily as well as to combine with the ERG system in a dark room. Without using a fundus camera, the OCT system provides en-face images from high-pass filtering and square of the OCT spectral signal for mouse retinal positioning in-vivo before acquiring ORG data. The 490nm LED for light stimulus is generated to make uniform illumination at the mouse retina using the Maxwellian view method. The common path of the OCT scanning light and the visible LED is built with achromatic doublet lens combinations based on optical simulation with Opticstudio® . The developed compact ORG system can not only observe light-evoked responses with 1~10 milliseconds but also be used for the studies of correlations between ORG and ERG in the mouse retina. © 2024 SPIE.

关键词： Optical tomography

Summon a Demon and Bind It: A Grounded Theory of LLM Red Teaming

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Inie, Nanna Stray, Jonathan Derczynski, Leon Paul G. Allen School of Computer Science & Engineering SeattleWA United States Department of Computer Science IT University of Copenhagen Copenhagen Denmark Berkeley Center for Human-Compatible AI UC Berkeley BerkeleyCA United States NVIDIA Corporation United States

Engaging in the deliberate generation of abnormal outputs from Large Language Models (LLMs) by attacking them is a novel human activity. This paper presents a thorough exposition of how and why people perform such attacks, defining LLM red-teaming based on extensive and diverse evidence. Using a formal qualitative methodology, we interviewed dozens of practitioners from a broad range of backgrounds, all contributors to this novel work of attempting to cause LLMs to fail. We focused on the research questions of defining LLM red teaming, uncovering the motivations and goals for performing the activity, and characterizing the strategies people use when attacking LLMs. Based on the data, LLM red teaming is defined as a limit-seeking, non-malicious, manual activity, which depends highly on a team-effort and an alchemist mindset. It is highly intrinsically motivated by curiosity, fun, and to some degrees by concerns for various harms of deploying LLMs. We identify a taxonomy of 12 strategies and 35 different techniques of attacking LLMs. These findings are presented as a comprehensive grounded theory of how and why people attack large language models: LLM red teaming. © 2023, CC BY.

关键词： Computational linguistics