Machine Learning is receiving growing attention due to recent successes in long standing problems such as speech recognition, computer vision and natural language processing. These successes are mainly due to a combination of advances in articial neural networks or deep learning methods, as well as the availability of large dataset and computational resources. This lecture covers the motivation for the underlying technology: When to use Machine Learning and particularly when to use Deep Learning? Why does Deep Learning work so well and what are its drawbacks? We will go over a number of examples that illustrate the modelling process, the design decisions and present a number of successful architectures.

Vlado Menkovski is an Assistant Professor of Data Mining at the Department of Mathematics and Computer Science, TU Eindhoven. You can find more information about him on his website.