Machine Unlearning - Addressing Bias, Privacy, and Regulation in LLMs and Multimodal Models

Explore machine unlearning techniques for large language models (LLMs) and multimodal models in this 35-minute conference talk from the Toronto Machine Learning Series. Led by Marija Stanojevic, Lead Applied Machine Learning Scientist at EudAImonia Science, delve into the critical aspects of adapting AI models to handle sensitive data across healthcare, finance, and personal information domains. Learn about cutting-edge advancements in machine unlearning dynamics, focusing on bias mitigation, data privacy protection, and legal compliance. Discover methods for identifying and removing unwanted data while maintaining model performance, and understand evaluation techniques that ensure successful data removal without compromising overall model behavior. Gain insights into how machine unlearning empowers stakeholders with data withdrawal capabilities, ultimately fostering trust in responsible AI development and deployment.