ASERG Research Seminar - Improving the Performance of Feature Selection

We are proud to announce the ASERG Research Seminar with Hakan Kiziloz, Research Fellow at Sheffield Hallam University.

Date: 20/10/2021, 14hrs BST.

Registration link (zoom link will be sent at the day of the event).

Title: Improving the Performance of Feature Selection

Presenter: Hakan Kiziloz (SHU)

Abstract:

With the advances in data storage and processing technologies, the amount of accessible data increases continuously. Data volume is useful; however, effective decision-making relies on the quality of available data. As data amount is beyond our manual processing capabilities, data mining tools along with machine learning techniques are used to extract meaningful information from data. Nevertheless, even these techniques may suffer from high amount of data, also known as the curse of dimensionality. Feature selection is one of the well-known and powerful preprocessing techniques to eliminate irrelevant/redundant parts of the data. We analyzed various strategies to improve the performance of feature selection task through six journal and two conference papers. These strategies include different evolutionary computation algorithms, classifier ensembles, various initial population generation mechanisms, and parallel computing. In this talk, first, I am going to give the formal definition of the feature selection problem. Then, I will briefly talk about the strategies we followed in the mentioned studies. Finally, I will provide an overview of possible future works on the feature selection task.