A Probabilistic Framework for Modular Continual Learning

By Lazar Valkov, Akash Srivastava, Swarat Chaudhuri, Charles Sutton. 2023

5 citations recorded by the source corpus.

Modular approaches that use a different composition of modules for each problem are a promising direction in continual learning (CL). However, searching through the large, discrete space of module compositions is challenging, especially because evaluating a composition's performance requires a round of neural network training. We address this challenge through a modular CL framework, PICLE, that uses a probabilistic model to cheaply compute the fitness of each composition, allowing PICLE to achieve both perceptual, few-shot and latent transfer. The model combines prior knowledge about good module compositions with dataset-specific information. We evaluate PICLE using two benchmark suites designed to assess different desiderata of CL techniques. Comparing to a wide range of approaches, we show that PICLE is the first modular CL algorithm to achieve perceptual, few-shot and latent transfer while scaling well to large search spaces, outperforming previous state-of-the-art modular CL approaches on long problem sequences.

Subfields: Artificial Intelligence

Citations in the corpus

• Recent Advances of Continual Learning in Computer Vision: An Overview (2021, neutral): “APD first uses a hierarchical knowledge consolidation method to construct a tree-like structure for the model parameters on the previous tasks where the root represents the parameters that are shared among all the tasks and the leaves represent the parameters that are specific to one task.”