This is not a stand-alone book but rather a collection of extensive references, specifically designed for senior machine learning researchers. It does not include many implementation details and instead focuses more on mathematical foundations.
This is a highly advanced book that strives to be as deep as possible, but it does so at the expense of details, implementation methods, and breadth of subtopics within subfields.
- (ML) Traditional Machine Learning Theories
- (DL) Deep Learning and Neural Network Basics
- (CV) Computer Vision Fundamentals
- (NLP) Natural Language Processing Basics
- (RL) Reinforcement Learning
- (RE) Recommendation Systems
- Advanced Deep Learning
- Semi-supervised Learning
- Self-supervised Learning
- Contrastive Learning
- Active Learning
- Continuous Learning
- Architecture Search
- Loss Function Theories
- Transformers
- Multi-modal Techniques
- AI simulation
- (LLM) Large Language Models
- Theory
- Prompt engineering
- RAG
- (GenAI) Generative AI
- Image Generation
- Chatbots
- Multi-modal cross application
- AI Agents
- (EAI) Embodied Intelligence
Nowadays, the application of AI nearly covers everything, but we still can trying to list all its ablities.
One way is categorized them by modalities of input/output:
- text to text
- natural chat
- text-based/code-based tasks (almost cover everything, but the intelligence level is limited)
- text to diagram: flow chart, mind map
- image to image: coloring, segementation
- text to image
- image to text
- image to structured data
Another way is categorized them by human-defined field:
