RanDumb: A Simple Approach that Questions the Efficacy of Continual Representation Learning.
Ameya PrabhuShiven SinhaPonnurangam KumaraguruPhilip H. S. TorrOzan SenerPuneet K. DokaniaPublished in: CoRR (2024)
Keyphrases
- learning process
- learning algorithm
- reinforcement learning
- prior knowledge
- hidden markov models
- subject matter
- knowledge acquisition
- learning scheme
- inductive learning
- answer questions
- neural network
- feature construction
- cognitive load
- incremental learning
- learning tasks
- learning systems
- image representation
- supervised learning
- active learning
- multiscale