Projected Latent Distillation for Data-Agnostic Consolidation in Distributed Continual Learning

TL;DR

This paper introduces Data-Agnostic Consolidation (DAC), a novel method for distributed continual learning that consolidates models without data access, enabling forward transfer and achieving state-of-the-art accuracy on multiple benchmarks.

Contribution

The paper proposes DAC, a double knowledge distillation approach using latent space projection, to enable data-agnostic model consolidation in distributed continual learning scenarios.

Findings

DAC achieves state-of-the-art accuracy on Split CIFAR100, CORe50, and Split TinyImageNet.

Even a single out-of-distribution image suffices for effective model consolidation.

DAC enables forward transfer between self-centered devices without sharing private data.

Abstract

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.