Meta-Learning: Theory, Algorithms and Applications

Deep neural networks (DNNs) with their dense and complex algorithms provide real possibilities for Artificial General Intelligence (AGI). Meta-learning with DNNs brings AGI much closer: artificial agents solving intelligent tasks that human beings can achieve, even transcending what they can achieve. Meta-Learning: Theory, Algorithms and Applications explains the fundamentals of meta-learning by providing answers to these questions: What is meta-learning?; why do we need meta-learning?; how are self-improved meta-learning mechanisms heading for AGI?; how can we use meta-learning in our approach to specific scenarios?

Format: Paperback / softback
Length: 402 pages
Publication date: 08 November 2022
Publisher: Elsevier Science & Technology

Deep neural networks (DNNs) with their intricate and dense algorithms hold immense potential for the development of Artificial General Intelligence (AGI). By leveraging meta-learning techniques, DNNs can significantly advance the pursuit of AGI, enabling artificial agents to perform tasks that surpass the capabilities of human beings.

Meta-Learning: Theory, Algorithms, and Applications, a comprehensive resource, delves into the realm of meta-learning and its application in conjunction with DNNs. It addresses crucial questions such as what meta-learning is, why it is necessary, how self-improved meta-learning mechanisms contribute to AGI, and how meta-learning can be applied to specific scenarios.

The book provides a comprehensive background on seven mainstream paradigms: meta-learning, few-shot learning, deep learning, transfer learning, machine learning, probabilistic modeling, and Bayesian inference. It then explores important state-of-the-art mechanisms and their variants for meta-learning, including memory-augmented neural networks, meta-networks, convolutional Siamese neural networks, matching networks, prototypical networks, relation networks, LSTM meta-learning, model-agnostic meta-learning, and the Reptile algorithm.

In a thorough exploration, the book delves into nearly 200 state-of-the-art meta-learning algorithms from top-tier conferences such as NeurIPS, ICML, CVPR, ACL, ICLR, and KDD. These algorithms are applied to 39 categories of tasks across 11 real-world application fields, including Computer Vision, Natural Language Processing, Meta-Reinforcement Learning, Healthcare, Finance and Economy, Construction Materials, Graphic Neural Networks, Program Synthesis, Smart City, Recommended Systems, and Climate Science. Each application field concludes with insights into future trends or by providing recommendations for future research.

By combining meta-learning with DNNs, the book offers a roadmap for advancing AGI and unlocking the full potential of artificial intelligence. It serves as a valuable resource for researchers, practitioners, and enthusiasts alike, providing a deep understanding of meta-learning and its transformative impact on the field of AI.

Weight: 858g
Dimension: 192 x 234 x 23 (mm)
ISBN-13: 9780323899314