For the November 7th IDiA/INVent Journal Club meeting, the IDiA lab is inviting prospective postdoc Aroof Aimen, a PhD Student at the Indian Institute of Technology, Ropar, India to present on some of her work.
Title: Improvising the generalizability of meta-learning approaches for few-shot learning.
Abstract: Deep learning models are powerful in extracting insights from data but demand abundant labeled examples, which can be limiting in real-world situations. To overcome this challenge, few-shot learning, a specialized field in machine learning, equips models trained on minimal data using a meta-learning technique. Meta-learning arranges training data into smaller tasks, enabling models to leverage past experiences when encountering new tasks similar to human learning. This talk will discuss assumptions concerning task distribution and label space and proposed solutions to enhance the resilience of meta-learning methods against distribution and label shifts.
Bio: Aroof Aimen is a Ph.D. student at the Indian Institute of Technology, Ropar, specializing in machine learning in the Computer Science and Engineering department. She has published her research work in top-tier machine learning conferences and journals like ECML, ICML, and Transactions in AI. During her Ph.D., Aimen completed a one-year research internship at Wadhwani AI, focusing on developing machine learning models to detect diseases from chest X-rays, particularly with limited data.