Accessories
Lectures
Seminar assignments
Exams / Izpiti
Additional material
Začetek predavanj/ The beginning of lectures:
Overview of the course & Introduction to Machine learning:
- overview of the course, student's obligations, grading
- ML and DM, KDD, Applications of ML
- classification and regression
- overview of ML methods (basic ML methods: decision/regression trees, NB/linear regression, K-NN, ANN, ) on the board
Exercises will start next week.
LECTURES:
- definition of learning, intelligence, artificial intelligence; why ML; video monkey
- natural learning; learning, intelligence and consciousness
- Basic ML principles (modeling, MDL, incremental, multiple explanations, estimating probabilities)
Začetek vaj/Beginning of exercises:
Uvod v R / An introduction to R
LECTURES:
- evaluating the results of ML:
- classification: accuracy, confution matrix, Brier, Information score, Sensitivity, specificity, ROC
- regression: MSE, RMSE, MAE, RMAE
- example of evaluating results of ML: breast cancer
- cross validation,
- Comparing algorithms, statistical tests, Bonferroni correction
LABS:
Data visualization
LECTURES:
- combining different algorithms: principle of multiple explanations; combining predictions;
- bagging, Random forests, boosting
- Estimating the attribute quality:
- classification: gain, gainR, 1-D, gini, ReliefF
- regression: Variance
- preprocessing: discretization, binarization, discrete into continuous
LABS:
- classification, CA, Sensitivity, Specificity, ROC, Brier Score
NO LECTURES THIS WEEK
NO LABS THIS WEEK
NO LECTURES, due to illness
LABS:
- classification models
- cross validation
LECTURES:
- unknown values,
- example of data preprocessing and binarization
- visualization
- Building and pruning decision and regresion trees;
- NB and its properties, KNN, locally weigted regression
LABS:
- classification models
- cross validation
LECTURES:
- Introduction to ANN
- (multi)layared feedforward ANN and (generalized.) delta learning rule
- overview of search algorithms for ML
LABS:
- regression
Reinforcement learning:
Natural language processing 1/2
LABS:
- classification exam problems
Natural language processing 2/2
Nature inspired computing
Heuristic search and two-player games
Agents and robots
LABS: textmining
Decision support