Efficient Feature Transformations for Discriminative and Generative Continual Learning

By Vinay Kumar Verma, Kevin J Liang, Nikhil Mehta, Piyush Rai, Lawrence Carin. 2021

93 citations recorded by the source corpus.

As neural networks are increasingly being applied to real-world applications, mechanisms to address distributional shift and sequential task learning without forgetting are critical. Methods incorporating network expansion have shown promise by naturally adding model capacity for learning new tasks while simultaneously avoiding catastrophic forgetting. However, the growth in the number of additional parameters of many of these types of methods can be computationally expensive at larger scales, at times prohibitively so. Instead, we propose a simple task-specific feature map transformation strategy for continual learning, which we call Efficient Feature Transformations (EFTs). These EFTs provide powerful flexibility for learning new tasks, achieved with minimal parameters added to the base architecture. We further propose a feature distance maximization strategy, which significantly improves task prediction in class incremental settings, without needing expensive generative models. We demonstrate the efficacy and efficiency of our method with an extensive set of experiments in discriminative (CIFAR-100 and ImageNet-1K) and generative (LSUN, CUB-200, Cats) sequences of tasks. Even with low single-digit parameter growth rates, EFTs can outperform many other continual learning methods in a wide range of settings.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning, 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 Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

In Meta & Continual Learning

Citations in the corpus

FeTrIL: Feature Translation for Exemplar-Free Class-Incremental Learning (2022, positive): “Our proposal is advantageous from memory and speed perspectives compared to mainstream methods [13, 33, 38, 42, 45, 47, 48, 49].”
Representation Compensation Networks for Continual Semantic Segmentation (2022, methodology): “Some other methods [45,57,75,76,82,91] increase the capacity of the model to have a better tradeoff of stability and plasticity, but with the cost of growing memory of the network.”
Convolutional Prompting meets Language Models for Continual Learning (2024, methodology): “(2) Dynamic Architecture-based methods learn new tasks by assigning distinct parameters for each task [13, 55– 57].”
Deep Class-Incremental Learning: A Survey (2023, background): “the generative model also suffers the catastrophic forgetting problem [126], [127], [128], which in turn increases the forgetting in incremental model training.”
A Metaverse: Taxonomy, Components, Applications, and Open Challenges (2022, unknown): “[318] proposed a simple task-specific feature map transformation strategy for continuous learning called Efficient Feature Transformations (EFT).”
Continual Contrastive Learning for Cross-Dataset Scene Classification (2022, background): “[14] demonstrate the contribution of their proposed Efficient Feature Transforms in generative models to overcome catastrophic forgetting.”