Advanced Methods for Optimizing and Accelerating Deep Learning Models

Häftad, Engelska, 2027

1 829 kr

Kommande

Beskrivning

Advanced Optimization and Acceleration Techniques for Deep Learning Models provides a comprehensive guide to enhancing deep learning models' efficiency, scalability, and performance, including large language models (LLMs). As AI systems grow in complexity, optimizing their training and deployment has become critical for achieving higher accuracy, faster inference, and reduced computational costs. This book explores cutting-edge optimization strategies, from gradient descent refinements and hyperparameter tuning to model compression, pruning, and hardware acceleration. AI is evolving rapidly, but existing deep learning resources often focus on building models rather than optimizing them for efficiency and scalability. As deep learning applications expand into cloud computing, edge AI, and real-time decision-making, a dedicated resource on optimization is essential. This book addresses this gap by providing a structured approach to making deep learning networks faster, more cost-effective, and more sustainable.

Explains the complexity problem in deep learning and explores optimization techniques tailored to different use cases
Training Efficiency: How to accelerate training without compromising accuracy using gradient descent optimizations, adaptive learning rates, and parallel processing
Model Deployment & Scalability: How to efficiently deploy deep learning models in cloud environments, edge devices, and mobile platforms using pruning, quantization, and distillation
Memory & Computational Constraints: How to reduce model size and inference latency for real-time applications using low-rank factorization, weight sharing, and model compression
Sustainability & Green AI: How to design energy-efficient AI systems that balance performance and resource consumption, making AI more accessible and cost-effective

Produktinformation

Utgivningsdatum:2027-02-01
Mått:191 x 235 x undefined mm
Vikt:450 g
Format:Häftad
Språk:Engelska
Antal sidor:200
Förlag:Elsevier Science
ISBN:9780443484957

Utforska kategorier

Programmeringsböcker inom Data och IT
Artificiell intelligens inom Data och IT
Systemvetenskap och AI inom Data och IT

Mer om författaren

Patrick Siarry was born in France in 1952. He received the Ph.D. degree in computer science and optimization from University Paris VI, Paris, France, in 1986, and the Doctorate of Sciences (Habilitation) degree in computer science and optimization from University Paris XI, Orsay, France, in 1994.,He was first involved in the development of analog and digital models of nuclear power plants with Electricité de France, Paris. Since 1995, he has been a Professor of Automatics and Informatics with Université Paris-Est Créteil, Créteil, France. His main research interests include computer-aided design of electronic circuits, cognitive intelligence, and the applications of new stochastic global optimization heuristics to various engineering fields, also including the fitting of process models to experimental data, the learning of fuzzy rule bases, and of neural networks. Rahma Fourati (Member, IEEE) received her PhD degree in computer engineering systems from the National Engineering School of Sfax (ENIS), Tunisia, with the Research Group in Intelligent Machines (REGIM), in 2021. She has been serving as an assistant professor at the Faculty of Law, Economics, and Management Sciences of Jendouba, Tunisia, since 2023. Additionally, she served as the chair of the IEEE Computational Intelligence Society for the period 2021-2022. Her research interests include affective computing, physiological signals, healthcare applications, deep neural network compression, and evolutionary computation. Dr. Jihene Tmamna received her Computer Science Engineering degree in 2017 and her Ph.D. in Computer Systems Engineering in 2023 from the National Engineering School of Sfax (ENIS), Tunisia. Her research interests include deep neural networks, neural network compression, and optimization algorithms. She has extensively worked on pruning and quantizing Convolutional Neural Network (CNN) architectures to develop Lightweight models. Additionally, she has applied pruning techniques to optimize multimodal CNN-based models in the context of the Audio-Visual Speech Enhancement (AVSE) challenge. She is currently a Postdoctoral Researcher at the University of Sfax, Tunisia. Asma Baghdadi holds a degree in Software Engineering and a Ph.D. in Information Sciences and Technologies. Since January 2022, she has served as an Assistant Professor at Esprit School of Business. Her research contributions shows her dedication to advancing knowledge at the intersection of technology and healthcare. As a researcher, she is affiliated with REGIM Lab, LISSI, and the AI4U research units.

Innehållsförteckning

1. Foundations of AI and Deep Learning Systems2. Optimization in Deep Learning: Motivation and Scope3. Neural Architecture Search (NAS) and Green AI4. Pruning Techniques for Model Compression5. Quantization for Efficient Inference6. Knowledge Distillation for Compact Models7. Sparsity and Efficient Architectures8. Hardware Acceleration for Deep Learning9. Federated Learning for Privacy-Preserving AI10. Split Learning for Collaborative Model Training11. Fog Computing for Edge AI