Course Program of  Soft Computing:

The course consists of 5 slots plus a conclidong laboratory:

1. Introduction of the Algorithmic Inference statistical framework to learn functions from data. The slot consists of 8 frontal lecture hours plus 8 lab hours. In the former we will recall the essential statistical tools allowing learning with computer programs. In the lab learning algorithms will be implemented to infer non linear functions from examples.

2. Cluster Analysis. The slot consists of 8 frontal lecture hours plus 12 lab hours. The former concerns the four ingredients of an efficient clustering of sparse data: i) decision function ii) loss function iii) data representation, and iv) template identification. In the lab various clustering algorithms will be implemented in relation to different data representation. As for thelatter Principal Component Analysis and Independent component Analysis will be employed.

3. Computazional Leraning. The slot consists of 8 frontal lecture hours plus 8 lab hours. The former concerns the learning of Boolean functions, and related feasibility and complexity results. In the lab, Support Vector Machines and decision trees will be tossed.

4. Neural networks. The slot consists of 8 frontal lecture hours plus 12 lab hours. The former, after considering a neural network taxonomy, extends to this framework results discussed in the previous modules. In the lab some widespread neural network models will be implemented, as well as a hybrid model implementing a mixed symbolic-subsymbolic learning.

5.  Learning fuzzy systems. The slot consists of 8 frontal lecture hours plus 12 lab hours. The former introduces fuzzy systems and main learning paradigms with this framework. In the lab a neurofuzzy system will be implemented.

The course ends with a set of mini-projects developed by small student teams under teacher supervision 4 hours.

See the web page of the laboratory of neural networks
LAREN(http://laren.dsi.unimi.it/apolloni/)
for details on the courses
and other activities