Projected Latent Distillation for Data-Agnostic Consolidation in Distributed Continual Learning
This addresses the challenge of knowledge consolidation in distributed edge learning for scenarios where devices are unwilling to share data, offering a practical solution for incremental improvements in privacy-preserving AI.
The paper tackles the problem of enabling forward transfer in distributed continual learning without accessing private data from self-centered devices, achieving state-of-the-art accuracy on benchmarks like Split CIFAR100, CORe50, and Split TinyImageNet, with results showing that even a single out-of-distribution image suffices for consolidation.
Distributed learning on the edge often comprises self-centered devices (SCD) which learn local tasks independently and are unwilling to contribute to the performance of other SDCs. How do we achieve forward transfer at zero cost for the single SCDs? We formalize this problem as a Distributed Continual Learning scenario, where SCD adapt to local tasks and a CL model consolidates the knowledge from the resulting stream of models without looking at the SCD's private data. Unfortunately, current CL methods are not directly applicable to this scenario. We propose Data-Agnostic Consolidation (DAC), a novel double knowledge distillation method that consolidates the stream of SC models without using the original data. DAC performs distillation in the latent space via a novel Projected Latent Distillation loss. Experimental results show that DAC enables forward transfer between SCDs and reaches state-of-the-art accuracy on Split CIFAR100, CORe50 and Split TinyImageNet, both in reharsal-free and distributed CL scenarios. Somewhat surprisingly, even a single out-of-distribution image is sufficient as the only source of data during consolidation.