Fundamental Principles and Methodologies in Artificial Intelligence

Daniela Iacoviello

The goal of artificial intelligence (AI) is to make machines able to take decisions and solve problems autonomously. This can pursued by reproducing processes like learning, recognition, choice and implementing generalization capabilities. AI is a vast discipline involving many topics, like planning, coordination and manipulation, knowledge representation, robotics, vision and speech ... and of course, machine learning (ML). Machine learning is considered nowadays one of the most intriguing approach of AI, able to face real applications too complex to be modelled in a classical way.

In this module the most common methods typical of ML are illustrated, discussing problems of classification, regression, clustering, dimensionality reduction, faced by means of supervisioned-non supervisioned learning and reinforcement learning. Problems regarding generalization and overfitting are considered, showing some useful strategies for an effective machine learning implementation; methods for assessing the reliability of the results, such as accuracy calculation or confusion matrix analysis, are illustrated. Implementation on real data are described.