42030 - Advanced Techniques in Machine Learning (TAVA) [UPC]

Type: S3 Course
Semester: Fall
ECTS: 6
Teaching Points: 15
Offer: Annual
Responsible Unit: CS-UPC
Responsible: Javier Béjar
Language: English
Requirements:

GOALS

This course is an extension of the methodologies learned from the previous machine learning courses. The main subjects treated in this couse include unsupervised learning from diferent perspectives (machine learning, intelligent data analysis, knowledge discovery) and case based reasoning.

CONTENTS

Part I – Unsupervised Learning

1. Unsupervised Learning and Data Mining
2. Elements of the process of unsupervised learning

Data representation
Data preprocessing and transformation
Distance functions

3. Unsupervised learning ouside machine learning

Cognitive Psicology
Numerical Taxonomy
Data analysis

4. Machine learning perspective

Conceptual Clustering
Concept formation
Evaluation of unsupervised learning
Sintatic Biasing, Semantic Biasing

5. Unsupervised Learning in Knowledge Discovery/Data Mining

Clustering Algorithms for Knoledge Discovery
Unsupervised Learning from patterns and structures
Association Rules, Time Series, Trees, Graphs

Part II – Case Based Learning and Reasoning

1. Introduction
2. CBR Foundations

Basic cicle of reasoning
Models for expertise representation

3. CBR Applications

Academic application: CHEF, CASEY, JULIA, HYPO, PROTOS
Practical application to complex domains
OPENCASE: A domain independent CBR system

4. Cases representation and organization

Representation structures
Library structures

5. CBR system phases

Case retrieval
Similarity evaluation
Adaptation strategies and/or methods
Learning

6. Reflective reasoning in CBR
7. Applications and tools for CBR developement

Industrial applicacions
Software tools

8. CBR evaluation
9. Advanced Research in CBR