1 semester

1.       

Title of the discipline (basic disciplines)

Machine learning

2.       

Course of Study

1 course, 1- 31 81 06 Web programming and Internet technologies

3.       

Semester of training
(for each semester a separate table)

1

4.       

Amount of credits
(academic plan)

5

5.       

Full name of the lecturer

Radyno Evgeniy Mefodievich

6.       

Objectives of studying the discipline

 

Familiarization of students with the main modern methods of automatic detection and description of regularities in data from the surrounding world.

Training methods of automating activities based on data analysis and modeling, without explicit pre-writing action algorithms.

 

As a result of studying the discipline, the trainee must know:

  • classification of data types available in the surrounding world;
  • formulation of regression, classification, ranking, distribution, diminution, clustering problems;
  • the main approaches to solving the problems of the above-mentioned regression problems, classification, ranking, distribution estimation, diminishing of size, clustering;

be able to:

  • determine the type of data available;
  • carry out primary data processing;
  • formalize the problem as a problem of regression, classification, ranking, evaluation of distribution, diminution of dimensions, clustering or their aggregate;
  • solve the mentioned problems with the help of modern software;
  • evaluate the quality of the proposed solution.

7.       

Prerequisites

 

Linear algebra, probability theory, mathematical statistics,

fundamentals of mathematical analysis, the fundamentals of functional analysis, the fundamentals of programming

8.       

Contents of the discipline

 

 

Types of variables (continuous, categorized, ordered). Precedent learning. Predictors and target variables. Problems of regression and classification. Probabilistic formulation of the classification problem. Steps of the analysis of data CRISP-DM. Retraining of the model and ways to combat the re-education of the model. Entropy and information. Density of distribution. The maximum likelihood method for estimating a parameter. Comparison of binary classifiers. ROC-curve. The loss function in assessing the quality of the model. Basic metrics for quality assessment.

Linear regression and regularization (LASSO, Ridge). Generalized linear regression. Logistic regression. Trees in the problem of classification and regression. The nearest-neighbor method. Support vector method. The principal component method. Composite models (Bagging). Neural networks (perceptron, LSTM). Clustering

9.       

Recommended literature

Basic

  1. Vapnik VN, Chervonenkis A.Ya. Theory of pattern recognition. Statistical learning problems. – Moscow, 1974. – 416 c.
  2. Trevor Hastie. Robert Tibshirani. Jerome Friedman. The Elements of. Statistical Learning. Data Mining, Inference, and Prediction. – Springer, 2009. – 745 p.

10.   

Teaching Methods

Lectures, laboratory practice, SSI (students scientific investigation)

11.   

Language of instruction

Russian

12.   

Conditions (requirements), routine monitoring

Laboratory work (when examining an examination, current academic performance is taken into account with a coefficient of 0.3)

13.   

Appraisal Form

exam