Transcript 幻灯片 1 - SJTU

Artificial Intelligence
Bo Yuan, Ph.D.
Professor
Shanghai Jiaotong University
Overview of Machine Intelligence
• Knowledge-based rules (expert system, automata, …)
– Symbolic representation in logics (Deep Blue)
• Kernel-based heuristics (MDA, PCA, SVM, …)
– Nonlinear connection for more representation (Neural Network)
• Inference (Bayesian, Markovian, …)
– To sparsely sample for convergence (GM)
• Interactive and stochastic computing (uncertainty,
heterogeneity)
– To possibly overcome the limit of Turin Machine
Interactions
The Framework to Study a System
Top-Down
Bottom-Up
How much can we represent and model a
complex and evolving network ?
Low Complexity Solutions for
High Complexity Problems
•
•
•
•
•
•
•
Convexity
Stability (Metastability)
Sampling
Ergodicity
Convergence
Regularization
Software and Hardware
Interactions
The Framework to Study a System
Top-Down
Bottom-Up
How much can we represent and model a
complex and evolving network ?
Data
Representation
Mathematical
Foundation
Graph
Mathematical
Representation
Typical
Algorithm
AI-Related
Question
Graph Theory and
Variable Reduction
Optimization
Liner Programming
Network Modularity
and Organization
Logic
Algebraic Logic
Random Boolean
Network, Automata
Network Structure
and Attractors
Circuit
Complex Number
and Control Theory
Linearization
Stability and control
Network Stability
and Control
Reasoning
Game Theory
Evolutionary Game
Nash Equilibrium
Markov Games
Inference
Bayes Theorem
Believe Propagation
Model Searching
Causality Inference
Discrete
Stochastic
Markov-based
Updating
Convergence
Meta-stability
Evolution and
Dynamics
Continuous
Stochastic
Stochastic
Differentials
Brownian integrals
Fokker-Planck
Network Dynamics
and Control
Review of Lecture One
• Overview of AI
–
–
–
–
Knowledge-based rules in logics (expert system, automata, …) : Symbolism in logics
Kernel-based heuristics (neural network, SVM, …) : Connection for nonlinearity
Learning and inference (Bayesian, Markovian, …) : To sparsely sample for convergence
Interactive and stochastic computing (Uncertainty, heterogeneity) : To overcome the
limit of Turin Machine
• Course Content
– Focus mainly on learning and inference
– Discuss current problems and research efforts
– Perception and behavior (vision, robotic, NLP, bionics …) not included
• Exam
– Papers (Nature, Science, Nature Review, Modern Review of Physics, PNAS, TICS)
– Course materials
Outline
•
•
•
•
•
Knowledge Representation
Searching and Logics
Perceiving and Acting
Learning
Uncertainty and Inference