P. Kiruthika / R. Leelavathi / V. Amala Deepa
Librería Samer Atenea
Kálamo Books
Librería Elías (Asturias)
Librería Kolima (Madrid)
Librería Proteo (Málaga)
This book is a research-oriented roadmap through modern ML, structured via three lenses: Research Perspectives, Recent Advances, and Future Directions. Part I establishes rigorous methodologies, data-centric AI (active learning, weak supervision, bias mitigation), and critical evaluation beyond single metrics. Part II surveys neural architectures (Transformers to State-Space models), optimization (SAM, double descent), generative modeling (diffusion unification), geometric deep learning, and self-supervised/multi-modal learning. Part III covers foundation models-scaling laws, emergent abilities (in-context learning, chain-of-thought), generative AI frontiers (hallucination, controllable generation), and neurosymbolic approaches (RAG, tool use). Part IV addresses grand challenges: efficient/sustainable ML, continual/meta-learning, interpretability/robustness/alignment (RLHF, causal representation), and pathways to AGI with societal impact. Each chapter includes open questions and practical research checklists, aiming to move readers from consumers to contributors.
