Generalized Variational Continual Learning

By Noel Loo, Siddharth Swaroop, Richard E. Turner. 2020

75 citations recorded by the source corpus.

Continual learning deals with training models on new tasks and datasets in an online fashion. One strand of research has used probabilistic regularization for continual learning, with two of the main approaches in this vein being Online Elastic Weight Consolidation (Online EWC) and Variational Continual Learning (VCL). VCL employs variational inference, which in other settings has been improved empirically by applying likelihood-tempering. We show that applying this modification to VCL recovers Online EWC as a limiting case, allowing for interpolation between the two approaches. We term the general algorithm Generalized VCL (GVCL). In order to mitigate the observed overpruning effect of VI, we take inspiration from a common multi-task architecture, neural networks with task-specific FiLM layers, and find that this addition leads to significant performance gains, specifically for variational methods. In the small-data regime, GVCL strongly outperforms existing baselines. In larger datasets, GVCL with FiLM layers outperforms or is competitive with existing baselines in terms of accuracy, whilst also providing significantly better calibration.

Subfields: Artificial Intelligence, Ocean Engineering, Probabilistic & Bayesian Methods, Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Ocean Engineering

In Probabilistic & Bayesian Methods

In Meta & Continual Learning

Citations in the corpus

A Comprehensive Survey of Continual Learning: Theory, Method and Application (2023, background): “lel branches (ACL [107], ReduNet [459], EPIE-Net [104]), adaptive layers (GVCL [276], DyTox [101]), masks or mask”
Continual Learning via Sequential Function-Space Variational Inference (2023, background): “…one task to another (Ahn et al., 2019; Aljundi et al., 2018; Chaudhry et al., 2018; Kirkpatrick et al., 2017; Lee et al., 2017; Liu et al., 2018; Loo et al., 2020; Nguyen et al., 2018; Park et al., 2019; Ritter et al., 2018; Schwarz et al., Proceedings of the 39 th International Conference on…”
DualPrompt: Complementary Prompting for Rehearsal-free Continual Learning (2022, methodology): “These methods can be further categorized as expanding the model [51, 65, 26, 31, 59], or dividing the model [35, 52, 58, 17, 11].”
Learning to Prompt for Continual Learning (2021, background): “The task-specific components can be identified by expanding the network [26, 31, 48, 50, 64, 68], or attend to task-specific sub-networks [19,35,51,59].”
Recent Advances of Continual Learning in Computer Vision: An Overview (2021, positive): “Loo et al. [30] proposed Generalized Variational Continual Learning which unifies both Laplacian approximation and variational approximation.”
Natural continual learning: success is a journey, not (just) a destination (2021, methodology): “…Bayesian continual learning algorithms include a hyperparameter λ that scales the prior compared to the likelihood term for the current task (Loo et al., 2020): To minimize this loss and thus ﬁnd a mode of the approximate posterior, it is common to employ pseudo-second-order stochastic…”
Dynamic VAEs with Generative Replay for Continual Zero-shot Learning (2021, neutral): “The second family of this field is regularization-based methods[17, 46, 25, 22, 1].”
Posterior Meta-Replay for Continual Learning (2021, background): “Examples include (Online) EWC [31, 64] and VCL [53].”
How to Reuse and Compose Knowledge for a Lifetime of Tasks: A Survey on Continual Learning and Functional Composition (2022, background): “Nguyen et al. (2018) developed the first variational continual learning (VCL) method, which, akin to EWC (Kirkpatrick et al., 2017) in the standard regularization-based setting, requires storing penalty terms for each network parameter.”