Privacy-Preserving Techniques in Generative AI and Large Language Models: A Narrative Review

Information, Journal Year: 2024, Volume and Issue: 15(11), P. 697 - 697

Published: Nov. 4, 2024

Generative AI, including large language models (LLMs), has transformed the paradigm of data generation and creative content, but this progress raises critical privacy concerns, especially when are trained on sensitive data. This review provides a comprehensive overview privacy-preserving techniques aimed at safeguarding in generative such as differential (DP), federated learning (FL), homomorphic encryption (HE), secure multi-party computation (SMPC). These mitigate risks like model inversion, leakage, membership inference attacks, which particularly relevant to LLMs. Additionally, explores emerging solutions, privacy-enhancing technologies post-quantum cryptography, future directions for enhancing AI systems. Recognizing that achieving absolute is mathematically impossible, emphasizes necessity aligning technical safeguards with legal regulatory frameworks ensure compliance protection laws. By discussing ethical implications underscores need balanced approach considers performance, scalability, preservation. The findings highlight ongoing research innovation develop keep pace scaling models, while adhering standards.

Language: Английский

---

Balancing Innovation and Privacy in the Age of Artificial Intelligence

IGI Global eBooks, Journal Year: 2025, Volume and Issue: unknown, P. 343 - 376

Published: Feb. 12, 2025

The rapid growth of artificial intelligence (AI) presents significant opportunities for innovation but also raises substantial privacy challenges. This chapter explores the intricate relationship between AI advancement and privacy, advocating a balanced approach that protects individual rights while fostering technological progress. It discusses AI's transformative potential in operational efficiency, personalization, predictive analytics, alongside concerns related to data dependency, security risks, algorithmic bias. reviews existing regulatory frameworks like GDPR emphasizes ethical guidelines focused on transparency accountability. proposes strategies such as privacy-preserving technologies synthetic reconcile with privacy. Finally, highlights need evolving laws public engagement ensure serves good without compromising rights.

Language: Английский

---

The Diagnostic Classification of the Pathological Image Using Computer Vision

Algorithms, Journal Year: 2025, Volume and Issue: 18(2), P. 96 - 96

Published: Feb. 8, 2025

Computer vision and artificial intelligence have revolutionized the field of pathological image analysis, enabling faster more accurate diagnostic classification. Deep learning architectures like convolutional neural networks (CNNs), shown superior performance in tasks such as classification, segmentation, object detection pathology. has significantly improved accuracy disease diagnosis healthcare. By leveraging advanced algorithms machine techniques, computer systems can analyze medical images with high precision, often matching or even surpassing human expert performance. In pathology, deep models been trained on large datasets annotated pathology to perform cancer diagnosis, grading, prognostication. While approaches show great promise challenges remain, including issues related model interpretability, reliability, generalization across diverse patient populations imaging settings.

Language: Английский

---

Who Will Author the Synthetic Texts? Evoking Multiple Personas from Large Language Models to Represent Users’ Associative Thesauri

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

---

CNN-Based Optimization for Fish Species Classification: Tackling Environmental Variability, Class Imbalance, and Real-Time Constraints

Information, Journal Year: 2025, Volume and Issue: 16(2), P. 154 - 154

Published: Feb. 19, 2025

Automated fish species classification is essential for marine biodiversity monitoring, fisheries management, and ecological research. However, challenges such as environmental variability, class imbalance, computational demands hinder the development of robust models. This study investigates effectiveness convolutional neural network (CNN)-based models hybrid approaches to address these challenges. Eight CNN architectures, including DenseNet121, MobileNetV2, Xception, were compared alongside traditional classifiers like support vector machines (SVMs) random forest. DenseNet121 achieved highest accuracy (90.2%), leveraging its superior feature extraction generalization capabilities, while MobileNetV2 balanced (83.57%) with efficiency, processing images in 0.07 s, making it ideal real-time deployment. Advanced preprocessing techniques, data augmentation, turbidity simulation, transfer learning, employed enhance dataset robustness imbalance. Hybrid combining CNNs intermediate improved interpretability. Optimization pruning quantization, reduced model size by 73.7%, enabling deployment on resource-constrained devices. Grad-CAM visualizations further enhanced interpretability identifying key image regions influencing predictions. highlights potential CNN-based scalable, interpretable classification, offering actionable insights sustainable management conservation.

Language: Английский

---

The Role of Synthetic Data in Robotics: Accelerating Development and Innovation

International Journal of Scientific Research in Computer Science Engineering and Information Technology, Journal Year: 2025, Volume and Issue: 11(1), P. 2991 - 2998

Published: Feb. 18, 2025

This comprehensive article explores the transformative role of synthetic data in modern robotics development and deployment. It examines how addresses fundamental challenges by providing artificially generated datasets that mimic real-world scenarios. The delves into core advantages data, including cost-effectiveness, scalability, risk mitigation robotic system development. analyzes major tools platforms used for generation, with detailed discussions CARLA, Gazebo, Unreal Engine. critical challenge reality gap between simulated real environments, exploring solutions through domain randomization sim-to-real transfer techniques. practical applications across autonomous driving, warehouse automation, surgery, demonstrating data's impact on these domains. Furthermore, investigates future directions, integration generative AI, automated scenario collaborative simulation insights continues to evolve shape

Language: Английский

---

Reproduction of Original Glioblastoma and Brain Metastasis Research Findings Using Synthetic Data

World Neurosurgery, Journal Year: 2025, Volume and Issue: 196, P. 123808 - 123808

Published: March 13, 2025

Language: Английский

---

Application of GenAI in Synthetic Data Generation in the Healthcare System

Published: Jan. 1, 2025

Language: Английский

---

A comprehensive survey of manual and dynamic approaches for cybersecurity taxonomy generation

Knowledge and Information Systems, Journal Year: 2025, Volume and Issue: unknown

Published: March 8, 2025

Language: Английский

---

Dynamic Graph Attention Meets Multi-Scale Temporal Memory: A Hybrid Framework for Photovoltaic Power Forecasting Under High Renewable Penetration

Processes, Journal Year: 2025, Volume and Issue: 13(3), P. 873 - 873

Published: March 16, 2025

In the context of accelerated global energy transition, power fluctuations caused by integration a high share renewable have emerged as critical challenge to security systems. The goal this research is improve accuracy and reliability short-term photovoltaic (PV) forecasting effectively modeling spatiotemporal coupling characteristics. To achieve this, we propose hybrid framework—GLSTM—combining graph attention (GAT) long memory (LSTM) networks. model utilizes dynamic adjacency matrix capture spatial correlations, along with multi-scale dilated convolution temporal dependencies, optimizes feature interactions through gated fusion unit. Experimental results demonstrate that GLSTM achieves RMSE values 2.3%, 3.5%, 3.9% for (1 h), medium-term (6 long-term (24 h) forecasting, respectively, mean absolute error (MAE) 3.8%, 6.2%, 7.0%, outperforming baseline models such LSTM, ST-GCN, Transformer reducing errors 10–25%. Ablation experiments validate effectiveness mechanism, 19% reduction in 1 h error. Robustness tests show remains stable under extreme weather conditions (RMSE 7.5%) data noise 8.2%). Explainability analysis reveals differentiated contributions features. proposed offers an efficient solution high-accuracy demonstrating its potential address key challenges integration.

Language: Английский

---