Exploiting Task Relationships for Continual Learning Using Transferability-Aware Task Embeddings
This work addresses the challenge of improving continual learning efficiency for deep neural network applications by exploiting task relationships, offering a practical and incremental advancement over existing strategies.
The paper tackles the problem of enhancing forward and backward transfer in continual learning by leveraging inter-task relationships, proposing a transferability-aware task embedding (H-embedding) and a hypernet framework, and achieves strong performance on benchmarks like CIFAR-100, ImageNet-R, and DomainNet compared to baseline and SOTA methods.
Continual learning (CL) has been a critical topic in contemporary deep neural network applications, where higher levels of both forward and backward transfer are desirable for an effective CL performance. Existing CL strategies primarily focus on task models, either by regularizing model updates or by separating task-specific and shared components, while often overlooking the potential of leveraging inter-task relationships to enhance transfer. To address this gap, we propose a transferability-aware task embedding, termed H-embedding, and construct a hypernet framework under its guidance to learn task-conditioned model weights for CL tasks. Specifically, H-embedding is derived from an information theoretic measure of transferability and is designed to be online and easy to compute. Our method is also characterized by notable practicality, requiring only the storage of a low-dimensional task embedding per task and supporting efficient end-to-end training. Extensive evaluations on benchmarks including CIFAR-100, ImageNet-R, and DomainNet show that our framework performs prominently compared to various baseline and SOTA approaches, demonstrating strong potential in capturing and utilizing intrinsic task relationships. Our code is publicly available at https://anonymous.4open.science/r/H-embedding_guided_hypernet/.