This subject introduces data-based machine learning techniques with
an important practical part that allows the student to get in touch with the reality of use and design of this kind of techniques.
The course covers the different design aspects of machine learning systems, from the details related to data entry, data analysis, obtaining main characteristics,
automatic grouping, the creation of patterns and models, and the generation of an automatic system that learn and recognize them. Finally, the evaluation of the behavior of the system when performing the task is studied assigned, estimation of its performance and effectiveness and the methods of adjusting it to optimize its
validity.
Another meaning of this subject is called “Machine Learning”. This subject has a lot in common with key subjects in the new Degree “Engineering and Data Systems” that began to be taught at ETSIT in the course 2020-21, specifically the subject “Machine Learning” that will be taught in the third year.
These systems are key in very diverse fields of application, from intelligent process automation to the analysis of large amounts of data to extract information and classify it into what has been called “Big Data”. The industry increasingly requires professionals with experience in learning automatic, automatic recognition of groups and patterns and extraction of information from data.
The areas of application are innumerable: identifying patterns in social media data, analyzing patterns of diseases and their diagnosis, optimization of decision-making, optimization of resources, analysis of customer behavior, operational organization, system failure detection, content analysis of written texts, image recognition, recognition of human activities etc.
Ultimately, data engineers will be able to clearly impact society in a multitude of
areas and in any sector of activity.
Introduction
1.1. Course overview
1.2. Data mining and machine learning
1.3. Simple Examples: Time, Iris, Labor Negotiations, Grain Sorting
1.4. Real areas of application: Website analysis, image analysis, disease diagnosis,
Marketing and Sales
- Classification and regression
2.1. Input data: Examples, attributes, classes
2.2. Classification, grouping and regression concept
2.3. ZeroR, 1-R, Bayes Rule and Naive Bayes
2.4. Decision trees and linear regression - PRACTICE 1: Using the WEKA tool
3.1. Introduction to Weka’s “explorer” interface
3.2. Data loading, data editor, data filtering
3.3. Display panel
3.4. Classification panel and examples - Evaluation
4.1. Training / assessment concept
4.2. Cross validation
4.3. Statistical significance
4.4. Over-training - PRACTICE 2: Examples of systems evaluation application
5.1. Application definition
5.2. Preparation of training and test data
5.3. Simple algorithm testing with simple data - Selection and transformation of characteristic features
6.1. Based on goodness estimate. Selection of main components. Supervised and non-supervised methods
supervised
6.2. Based on classification. Wrapper-based methods - PRACTICE 3: Study and selection of traits to improve the system
7.1. Try different attributes
7.2. Manual selection of attributes
7.3. Automatic selection of attributes: Selection of main components - Advanced sorting and grouping
8.1. Logistic regression, SVM. Competitive project presentation
8.2. K Nearest Neighbors Rule (K-NN), Joint Learning
8.3. Multilayer Perceptron (ANN)
8.4. Clustering techniques, Simple kmeans, Estimate-Maximize (EM) - INTRODUCTION TO PRACTICE 4
9.1. Tutorial on different data available
9.2. Tutorial on possible algorithms to use in each problem - PRACTICE 4. Development of an evolved learning and classification system
