Speaker: Han Liu

The first section of this presentation will explore the motivation of research in machine learning as well as identify its main differences to data mining from both philosophical and practical viewpoints. Some popular applications of machine learning will be listed. The presenter will also introduce three types of logic widely used in practice namely, deterministic logic, probabilistic logic and fuzzy logic.

The second section will focus on introducing some popular machine learning methods. For each of the methods, the presenter will deliver explanations in the context of both human learning and machine learning. The learning methods are also judged regarding whether they are also good at data mining tasks.

The third section will introduce some ways to evaluate machine learning algorithms in academic aspects and learned models in practical aspects. In particular, this section will introduce two testing methods namely isolation method and integral method, which are used for the estimation of accuracy in data mining and machine learning tasks respectively. The presenter will also introduce the ways to validate for computational efficiency in the two subject areas.

The fourth section will explore the nature of existing issues and proposed solutions to suggest the further directions of research in data mining and machine learning. In detail, the presenter will introduce a framework for control of machine learning tasks, which is recently proposed for future research. The presenter will also explore the view that the proposed framework would involve interdisciplinary research in order to comprehensively fill the existing gaps .