CMPS 242 Midterm Topics (Fall 2005)

(Version 1)

The midterm will be in class Tuesday Feb. 22.  The exam will be closed-book, although students may have one 3x5 card of handwritten notes (both sides).   Here are the topics we have covered so far in CMPS 242 this quarter:

 

  1. What is Machine Learning
  2. Classification and Concept learning, hypothesis class and version spaces
  3. VC (Vapnik-Chervonenkis) dimension and shattering points by hypothesis class
  4. Perceptron algorithm and Widrow-Hoff gradient descent techniques for learning halfspaces
  5. Logistic regression
  6. Linear discriminant analysis
  7. Noise and causes (label errors, attribute errors, features or hypothesis class may not fit phenomina exactly)
  8. Overfitting – Using test sets to detect reduction in generalization
  9. Basic probability (sample space, events, random variables, independence, conditional probability, Bayes rule)
  10. Estimating probabilities (e.g. the bias of a coin flip): maximum likelihood, priors, maximum aÕposteriori, mean aÕposteriori, and Laplacian estimates of probability
  11. Hypotheses as models generating the data, identifying the maximum likelihood hypothesis, the maximum aÕposteriori hypothesis, and the mean aÕposteriori predictions
  12. Least squares regression as Maximum likelihood (assuming Gaussian noise)
  13. Asymmetric losses, Bayes risk, and Bayes optimal predictions
  14. Decision boundaries and discriminant functions
  15. Learning Gaussian distributions – biased and unbiased estimators of the variance
  16. Gaussians with same co-variance matrices give linear decision boundaries
  17. Bayes Nets (graphical models, belief networks)
  18. Na•ve Bayes algorithm
  19. Decision Trees:  Greedy construction of trees, information gain criterion, applying a split criterion (impurity function, e.g. information gain) to select tests at nodes, overfitting and pruning
  20. Feed forward artificial neural networks, training using backprop algorithm (gradient descent)
  21. Nearest Neighbor algorithm and variants.  Instance based density estimation.
  22. Support vector machines and Kernel functions
  23. Boosting

 

Recommended Problems: