Süreyya Özöğür
METU, Institute of Applied Mathematics
The Analysis of Patterns and Foundations of Computational Statistics
A subfield of the artificial intelligence, machine learning is concerned with the development of algorithms which allow computers to “learn”. It is a process of training the system with a massive set of examples, extracting rules and patterns to be able to make a prediction on the test example. Some kinds of machine learning techniques are related with “data mining” and heavily concerned with the statistics by focusing on computational complexity. Common machine learning algorithm types include supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, etc.. There are different kinds of applications in this field including natural language processing, search engines, medical diagnosis, bioinformatics, detecting credit card fraud, stock market analysis, classifying DNA sequences, speech and handwriting recognition, and object recognition in computer vision. In the first part of this talk, after giving some basic introduction to one of the powerful methods in supervised learning, Support Vector Machines (SVMs), its application to pattern analysis of eukaryotic pro-peptide cleavage sites will be discussed. In the second part, I will give introduction into the mathematical foundations of learning theory to introduce probabilistic error estimates on given examples while assessing the quality of the predictions. The introduction of further modern methods of applied mathematics is a scientific challenge, but it can become very worthwhile.