A Review of Large Language Models: Fundamental Architectures, Key Technological Evolutions, Interdisciplinary Technologies Integration, Optimization and Compression Techniques, Applications, and Challenges DOI Open Access
Songyue Han, Mingyu Wang, Jialong Zhang

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(24), P. 5040 - 5040

Published: Dec. 21, 2024

Large language model-related technologies have shown astonishing potential in tasks such as machine translation, text generation, logical reasoning, task planning, and multimodal alignment. Consequently, their applications continuously expanded from natural processing to computer vision, scientific computing, other vertical industry fields. This rapid surge research work a short period poses significant challenges for researchers comprehensively grasp the dynamics, understand key technologies, develop field. To address this, this paper provides comprehensive review of on large models. First, it organizes reviews background current status, clarifying definition models both Chinese English communities. Second, analyzes mainstream infrastructure briefly introduces optimization methods that support them. Then, conducts detailed intersections between interdisciplinary contrastive learning, knowledge enhancement, retrieval hallucination dissolution, recommendation systems, reinforcement models, agents, pointing out valuable ideas. Finally, deployment identifies limitations they face, an outlook future directions. Our aims not only provide systematic but also focus integration with hoping ideas inspiration carry secondary development

Language: Английский

Who Will Author the Synthetic Texts? Evoking Multiple Personas from Large Language Models to Represent Users’ Associative Thesauri DOI Creative Commons
Maxim Bakaev,

Svetlana Gorovaia,

Olga А. Mitrofanova

et al.

Big Data and Cognitive Computing, Journal Year: 2025, Volume and Issue: 9(2), P. 46 - 46

Published: Feb. 18, 2025

Previously, it was suggested that the “persona-driven” approach can contribute to producing sufficiently diverse synthetic training data for Large Language Models (LLMs) are currently about run out of real natural language texts. In our paper, we explore whether personas evoked from LLMs through HCI-style descriptions could indeed imitate human-like differences in authorship. For this end, ran an associative experiment with 50 human participants and four artificial popular LLM-based services: GPT-4(o) YandexGPT Pro. each five stimuli words selected university websites’ homepages, asked both groups subjects come up 10 short associations (in Russian). We then used cosine similarity Mahalanobis distance measure between association lists produced by different humans personas. While difference significant associators gender age groups, neither case ChatGPT or YandexGPT. Our findings suggest services so far fall at imitating thesauri authors. The outcome study might be interest computer linguists, as well AI/ML scientists prompt engineers.

Language: Английский

Citations

0

Distinguishing Reality from AI: Approaches for Detecting Synthetic Content DOI Creative Commons

David Ghiurău,

Daniela Elena Popescu

Computers, Journal Year: 2024, Volume and Issue: 14(1), P. 1 - 1

Published: Dec. 24, 2024

The advancement of artificial intelligence (AI) technologies, including generative pre-trained transformers (GPTs) and models for text, image, audio, video creation, has revolutionized content generation, creating unprecedented opportunities critical challenges. This paper systematically examines the characteristics, methodologies, challenges associated with detecting synthetic across multiple modalities, to safeguard digital authenticity integrity. Key detection approaches reviewed include stylometric analysis, watermarking, pixel prediction techniques, dual-stream networks, machine learning models, blockchain, hybrid approaches, highlighting their strengths limitations, as well accuracy, independent accuracy 80% analysis up 92% using modalities in approaches. effectiveness these techniques is explored diverse contexts, from identifying deepfakes media AI-generated scientific texts. Ethical concerns, such privacy violations, algorithmic bias, false positives, overreliance on automated systems, are also critically discussed. Furthermore, addresses legal regulatory frameworks, intellectual property emerging legislation, emphasizing need robust governance mitigate misuse. Real-world examples systems analyzed provide practical insights into implementation Future directions developing generalizable adaptive fostering collaboration between stakeholders, integrating ethical safeguards. By presenting a comprehensive overview AIGC detection, this aims inform researchers, policymakers, practitioners addressing dual-edged implications AI-driven creation.

Language: Английский

Citations

3

Exploring the Next Frontier in Wireless Communication: 5G and Beyond for Enhanced Reliability and Low Latency in IoT and Autonomous Technologies DOI Open Access
Deepak Kumar Sharma, K. Lakshmi Narayana,

P Shyamala

et al.

Nanotechnology Perceptions, Journal Year: 2024, Volume and Issue: unknown, P. 676 - 689

Published: Dec. 1, 2024

This research focuses on how 5G and beyond technologies might be the game changers in reliability, low latency, efficiency, improvement of IoT autonomous systems, such as electric vehicles. It addresses advancements 6G-based communication networks integrated with machine learning edge computing to enhance vehicle performance, energy management, vehicle-to-infrastructure (V2I) communication. Extensive experimentation conducted greatly led discovery important improvements response time. Latency was reduced by much 45 per cent when compared 4G networks, this meant that 6G enabled potential increases up 60 over data throughput reliability high-density environments. In addition that, AI application towards predictive maintenance battery optimization an increase 30 for applications intelligence a more sustainable EV system. The results further reveal promise AI-based security ML-based 25% reduction network vulnerabilities traditional protocols. inform transformative capability next generations fulfil their scope remodelling future vehicles systems. Future will focus overcoming present infrastructure deficiencies improving algorithms behind real-time decision-making processes support scalable, energy-efficient, secure ecosystems.

Language: Английский

Citations

0

Multimodal Large Language Model-Based Fault Detection and Diagnosis in Context of Industry 4.0 DOI Open Access
Khalid Alsaif, Aiiad Albeshri, Maher Khemakhem

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(24), P. 4912 - 4912

Published: Dec. 12, 2024

In this paper, a novel multimodal large language model-based fault detection and diagnosis framework that addresses the limitations of traditional approaches is proposed. The proposed leverages Generative Pre-trained Transformer-4-Preview model to improve its scalability, generalizability, efficiency in handling complex systems various scenarios. Moreover, synthetic datasets generated via models augment knowledge base enhance accuracy imbalanced framework, hybrid architecture integrates online offline processing, combining real-time data streams with fine-tuned for dynamic, accurate, context-aware suited industrial settings, particularly focusing on security concerns, introduced. This comprehensive approach aims address challenges advance field toward more adaptive efficient systems. paper presents detailed literature review, including taxonomy methods their applications across domains. study discusses case results comparisons, exploring implications future developments within Industry 4.0 technologies.

Language: Английский

Citations

0

A Review of Large Language Models: Fundamental Architectures, Key Technological Evolutions, Interdisciplinary Technologies Integration, Optimization and Compression Techniques, Applications, and Challenges DOI Open Access
Songyue Han, Mingyu Wang, Jialong Zhang

et al.

Electronics, Journal Year: 2024, Volume and Issue: 13(24), P. 5040 - 5040

Published: Dec. 21, 2024

Large language model-related technologies have shown astonishing potential in tasks such as machine translation, text generation, logical reasoning, task planning, and multimodal alignment. Consequently, their applications continuously expanded from natural processing to computer vision, scientific computing, other vertical industry fields. This rapid surge research work a short period poses significant challenges for researchers comprehensively grasp the dynamics, understand key technologies, develop field. To address this, this paper provides comprehensive review of on large models. First, it organizes reviews background current status, clarifying definition models both Chinese English communities. Second, analyzes mainstream infrastructure briefly introduces optimization methods that support them. Then, conducts detailed intersections between interdisciplinary contrastive learning, knowledge enhancement, retrieval hallucination dissolution, recommendation systems, reinforcement models, agents, pointing out valuable ideas. Finally, deployment identifies limitations they face, an outlook future directions. Our aims not only provide systematic but also focus integration with hoping ideas inspiration carry secondary development

Language: Английский

Citations

0