Enhancing Continual Learning with IMEX-Reg: A Robust Approach to Mitigate Catastrophic Forgetting

Systems face challenges with continual learning and suffer from catastrophic forgetting. IMEX-Reg, a novel framework, enhances continual learning by combining contrastive representation learning with consistency regularization. It outperforms existing methods, improves task accuracy, and demonstrates resilience against disturbances.