Continual learning - the impact of task similarity
Neural networks project on continual learning, investigating the impact of task similarity on the performance of neural networks in a continual learning setting. With Lorenzo Bardone and Harshith Gowrachari.
This project studies continual learning in neural networks, with a focus on how task similarity affects performance when models are trained sequentially on multiple tasks. Together with Lorenzo Bardone and Harshith Gowrachari, we explored when shared structure between tasks can reduce forgetting and improve transfer across the learning process.