1. Overview
   • Attempts to cover key techniques compactly
   • Idiosyncratic, but nothing else available
   • Derivations helpful for math. types; intuition other ways
   • Techniques do get applied without explanation
   
   
2. Chapters
   • Chapter 2: Information Theory, Classification, MDL
   • Chapter 3: State Spaces and Search
   • Chapter 4: Vectors, Eigenvalues, PCA
   • Chapter 5: Dynamical Systems, Eigenvalues again, Non-linear Systems
   • Chapter 6: Control and Optimization, Continuous and Discrete
   
   
3. Chapter 2: Inf. Theory, Classification, MDL
   • Information Theory - used in Problem 1
   • Classification - like standard statistics
   • Minimum Description Length
   • Used in compression; also Language Learning
   
   
4. Chapter 3: State Spaces and Search
   • State Spaces a very general methodology
   • Discrete Search a big part of symbolic AI
   • Two-Person games and Alpha-Beta pruning
   • Also Games against Nature and Expected Values
   • Used in Reinforcement Learning and Weight Space
   
   
5. Chapter 4: Vectors, Eigenvalues, PCA
   • Data Compression as an ubiquitous issue - 2 approaches here
   • Eigenvectors as a general technique
   • PCA : Eigenvectors of the Covariance Matrix
   • Very large spaces
   • Clustering as an alternative data compression technique
   • MDL can be considered as the general case
   
   
6. Chapter 5: Dynamical Systems
   • General Study of how systems evolve through time
   • Differential Equations or discrete approximations
   • Eigenvectors employed here, too.
   • Applied in Hopfield Nets and in Weight Space search
   
   
7. Chapter 6: Control and Optimization
   • Optimization another ubiquitous issue
   • Ballard focuses on Control of dynamical system
   • Continuous and Discrete Techniques