CCL

Összesen 3 találat.
#/oldal:
Részletezés:
Rendezés:

1.

001-es BibID:BIBFORM128457
Első szerző:Lakatos Róbert (informatikus)
Cím:Investigating the Performance of Retrieval-Augmented Generation and Domain-Specific Fine-Tuning for the Development of AI-Driven Knowledge-Based Systems / Lakatos Róbert, Pollner Péter, Hajdu András, Joó Tamás
Dátum:2025
ISSN:2504-4990
Megjegyzések:Generative large language models (LLMs) have revolutionized the development of knowledge-based systems, enabling new possibilities in applications like ChatGPT, Bing, and Gemini. Two key strategies for domain adaptation in these systems are Domain-Specific Fine-Tuning (DFT) and Retrieval-Augmented Generation (RAG). In this study, we evaluate the performance of RAG and DFT on several LLM architectures, including GPT-J-6B, OPT-6.7B, LLaMA, and LLaMA-2. We use the ROUGE, BLEU, and METEOR scores to evaluate the performance of the models. We also measure the performance of the models with our own designed cosine similarity-based Coverage Score (CS). Our results, based on experiments across multiple datasets, show that RAG-based systems consistently outperform those fine-tuned with DFT. Specifically, RAG models outperform DFT by an average of 17% in ROUGE, 13% in BLEU, and 36% in CS. At the same time, DFT achieves only a modest advantage in METEOR, suggesting slightly better creative capabilities. We also highlight the challenges of integrating RAG with DFT, as such integration can lead to performance degradation. Furthermore, we propose a simplified RAG-based architecture that maximizes efficiency and reduces hallucination, underscoring the advantages of RAG in building reliable, domain-adapted knowledge systems.
Tárgyszavak:Műszaki tudományok Informatikai tudományok idegen nyelvű folyóiratközlemény külföldi lapban
folyóiratcikk
generative large language model
domain-specific fine-tuning
knowledge-based system
natural language processing
machine and deep learning
Megjelenés:Machine Learning and Knowledge Extraction. - 7 : 1 (2025),p. 1-18. -
További szerzők:Pollner Péter Hajdu András (1973-) (matematikus, informatikus) Joó Tamás
Pályázati támogatás:RRF-2.3.1-21-2022-00006
Egyéb
KDP-2021
Egyéb
Internet cím:Szerző által megadott URL
DOI
Intézményi repozitóriumban (DEA) tárolt változat
Borító:

2.

001-es BibID:BIBFORM118938
035-os BibID:(Scopus)85187896696 (WoS)001182558700001
Első szerző:Lakatos Róbert (informatikus)
Cím:A multimodal deep learning architecture for smoking detection with a small data approach / Róbert Lakatos, Péter Pollner, András Hajdu, Tamás Joó
Dátum:2024
ISSN:2624-8212
Megjegyzések:Covert tobacco advertisements often raise regulatory measures. This paper presents that artificial intelligence, particularly deep learning, has great potential for detecting hidden advertising and allows unbiased, reproducible, and fair quantification of tobacco-related media content. We propose an integrated text and image processing model based on deep learning, generative methods, and human reinforcement, which can detect smoking cases in both textual and visual formats, even with little available training data. Our model can achieve 74% accuracy for images and 98% for text. Furthermore, our system integrates the possibility of expert intervention in the form of human reinforcement. Using the pre-trained multimodal, image, and text processing models available through deep learning makes it possible to detect smoking in different media even with few training data.
Tárgyszavak:Természettudományok Matematika- és számítástudományok idegen nyelvű folyóiratközlemény külföldi lapban
folyóiratcikk
AI supported preventive healthcare
pre-training with generative AI
multimodal deep learning
automated assessment of covert advertisement
few-shot learning
smoking detections
artificial intelligence
Megjelenés:Frontiers in Artificial Intelligence. - 7 (2024), p. 1-8. -
További szerzők:Pollner Péter Hajdu András (1973-) (matematikus, informatikus) Joó Tamás
Pályázati támogatás:GINOP-2.3.2-15-2016-00005
GINOP
TKP2021-NKTA-34
Egyéb
KDP-2021
Egyéb
RRF-2.3.1-21-2022-00006
Egyéb
Internet cím:Szerző által megadott URL
DOI
Intézményi repozitóriumban (DEA) tárolt változat
Borító:

3.

001-es BibID:BIBFORM128461
035-os BibID:(WOS)001414387500030 (Scopus)85216389262
Első szerző:Pándy Árpád (szoftverfejlesztő rendszermérnök)
Cím:Investigating the Influence of Hyperparameters on the Optimal Time-Series Prediction Ability of Generative Large Language Models / Arpad Pandy, Róbert Lakatos, Péter Pollner, Eszter Csernai, András Hajdu
Dátum:2024
Megjegyzések:This study investigates the optimal hyperparameter settings for Large Language Models (LLMs) in predicting time-series data. We conducted extensive experiments using both synthetic datasets and real agricultural data to evaluate the performance of several LLMs. Our findings reveal that the predictive accuracy of these models is highly dependent on the configuration of the temperature and Top- p hyperparameters. Specifically, models demonstrated optimal performance when the ratio between these two parameters was maintained below a threshold, with the most effective ratios being under 2.0 for Mistral-small and around 4.0 for Gemma-7B. Furthermore, Mistral-small showed lower variability in performance across different settings, indicating higher determinism. By aggregating data from multiple experimental conditions, we identified that maintaining a temperature to Top-p ratio below 1.36 covers 50% of top-performing pairs for Mistral-small and below 2.75 for Gemma-7B. If we consider only hyperparameters, that has such ratios, we can decrease the MSE value by 32% on average. These insights provide a foundation for enhancing the accuracy of LLMs in time-series forecasting, offering practical guidelines for hyperparameter optimization. Our results underscore the potential of LLMs to outperform traditional methods, particularly in datasets with periodic patterns, thereby contributing to the growing literature on the application of generative models in time-series analysis.
ISBN:9798350387889
Tárgyszavak:Műszaki tudományok Informatikai tudományok előadáskivonat
könyvrészlet
Temperature distribution
Temperature dependence
Accuracy
Large language models
Time series analysis
Predictive models
Data models
Forecasting
Tuning
Synthetic data
generative large language models
hyperparam-eters optimization
time-series prediction
synthetic and agricul-tural data
Megjelenés:2024 IEEE 3rd Conference on Information Technology and Data Science (CITDS). - p. 158-163. -
További szerzők:Lakatos Róbert (1986-) (informatikus) Pollner Péter Csernai Eszter Hajdu András (1973-) (matematikus, informatikus)
Pályázati támogatás:2020-1.1.2-PIACI-KFI-2021-00223
Egyéb
TKP2021-NKTA-34
Egyéb
Internet cím:Szerző által megadott URL
DOI
Intézményi repozitóriumban (DEA) tárolt változat
Borító:
Rekordok letöltése1