590 Course Seminar - Batuhan Edgüer - Meta Modeling For AI Solutions Based on Existing AI Configurations

DEPARTMENT OF COMPUTER ENGINEERING

CSE 590 COURSE SEMINAR

Meta Modeling For AI Solutions Based on Existing AI Configurations

Batuhan Edgüer

Date & Time: Monday, January 27, 2025, 16:00

Place: https://meet.google.com/ovx-cpww-ath

All interested are cordially invited.

ABSTRACT

Rapid advancements in AI/ML research have made it challenging to identify suitable models due to the volume of research, lack of structured metadata, and resource-heavy experimentation. This paper develops a meta-learning-based recommendation system to automate model selection based on task-specific needs. Using meta-learning techniques and a PostgreSQL database, the system analyzes metadata from AI/ML papers, identifies task-model patterns, and provides real-time recommendations. It reduces time and computational costs, offering a scalable and self-improving framework for AI model recommendations.

BIOGRAPHY

Batuhan Edgüer was born on March 10, 1995, in Ankara, Turkey. He earned a Bachelor of Science degree in Biomedical Engineering from Yeditepe University in 2020, where he completed an engineering project on Tumor Segmentation. Currently, he is pursuing a Master of Science in Computer Science Engineering at Yeditepe University. Batuhan has experience as a Teaching Assistant and Software Engineer, where he contributed to projects involving medical imaging and software development. His technical expertise includes Python, C++, and CUDA, and he has worked on diverse projects, including Bitcoin price prediction using sentiment analysis and tumor segmentation. He participated in the "Karaciğer Algoritmaları Yarışıyor" competition held in Izmir, Turkey, on April 26, 2018. Beyond academia, he enjoys board games, trading card games, and fantastic literature.