El objetivo principal de la asignatura es que el alumno sea capaz de aplicar las técnicas clásicas más importantes de Reconocimiento de Patrones y las Redes Neuronales Artificiales más extendidas (tanto supervisadas como no supervisadas) a problemas concretos, algunas veces con datos sintéticos y otras veces con datos del mundo real (p.e. imágenes) utilizando Matlab para ello. El alumno debe comprender las características principales comunes a todas las técnicas basadas en el aprendizaje, así como a evaluar las ventajas e inconvenientes de la utilización de cada técnica en particular para un problema en concreto. Para ello van resolviendo los mismos problemas mediante ejercicios planteados con las distintas técnicas explicadas durante el curso y realizan igualemente 2 trabajos que engloban el análisis comparativo de las técnicas explicadas.
Pascual Campoy
Departamento de Automática, Electrónica e Informática Industrial.
Escuela Técnica Superior de Ingenieros Industriales
Asignatura Fundamental de Postgrado en el Master en Automática y Robótica (con acreditación al programa de calidad).
Última revisión: noviembre 2009.
Recognition of visual patterns.
Credits: 3 ECTS.
REQUIRED PREVIOUS SUBJETCS
- General Algebra
- Differential equations
DESCRIPCIÓN GENERAL DE LA ASIGNATURA
The main objective is that the student can apply the most important techniques for Machine Learning, both the ?Classical Techniques? and those based on ?Artificial Neural Networks?, to solve problems using actual data, some of them based on synthetic data, useful for getting familiar with the techniques, and some others based on data from real-word applications. The problems include both supervised learning problems, as well as unsupervised problems. The student is aimed to understand the features common to any kind of machine learning technique, and also to be able to understand the advantages and drawbacks of every technique in order to solve a particular problem. The classical techniques are studied as the reference techniques that used mathematical solutions and with which the new soft-computing techniques based on Neural Networks are to be compared with. The examples are solved using Matlab © and the specific toolbox of Statistics and Neural-Networks. A good motivation for using the techniques based in Neural Networks is given, by presenting the main features and the general methodology of such bio-inspired techniques, when compared to classical ones.
GENERAL OBJETIVES
- Ability to apply theoretical concepts and techniques included in the table of contents.
- Ability to solve practical problems with both synthetic data and real word data using Matlab ©.
- Ability to team working.
EVALUATION ACTIVITIES
- Classroom exercises.
- Two collaborative Homeworks.
- Individual exam.
A possible relative weightiness for global marking is 2, 4 and 4..
Escribir un comentario