March 25, 2022

About the speaker

Md Imrul Kaish has been a graduate student at the Department of Electrical Engineering since Spring'20. Hopefully, he will graduate in the spring'22 and begin looking for a job as a software engineer. He served as a lecturer at Daffodil International University, a famous private university in Bangladesh, prior to joining UTRGV. He graduated from the Islamic University of Technology in Dhaka, Bangladesh, with a bachelor's degree in electrical engineering. Under the supervision of Dr. Jia Chen, who recently moved to UCR, he published a paper on discriminative analysis of COVID-19 symptoms dataset. He is an avid learner who wishes to pursue a career in AI-related disciplines.

Feature selection and data-centric AI

Feature selection is a dimensionality reduction strategy that allows to choose a small number of features while removing the ones that aren't necessary for learning a model. Novel discriminative analysis method (DNA) was used to choose features efficiently, and its performance was assessed in a supervised and unsupervised learning setting. Data-centric AI is a new trend led by Andrew Ng that recommends shifting the focus of learning model development from model-centric to data-centric. To improve data quality while using the same model. It has several major advantages, including faster model learning, shorter deployment times, and increased accuracy.