Subset Replay Based Continual Learning for Scalable Improvement of Autonomous Systems

By Pratik Prabhanjan Brahma, Adrienne Othon. 2018

19 citations recorded by the source corpus.

While machine learning techniques have come a long way in showing astounding performance on various vision problems, the conventional way of training is not applicable for learning from a sequence of new data or tasks. For most real life applications like perception for autonomous vehicles, multiple stages of data collection are necessary to improve the performance of machine learning models over time. The newer observations may have a different distribution than the older ones and thus a simply fine-tuned model often overfits while forgetting the knowledge from past experiences. Recently, few lifelong or continual learning approaches have shown promising results towards overcoming this problem of catastrophic forgetting. In this work, we show that carefully choosing a small subset of the older data with the objective of promoting representativeness and diversity can also help in learning continuously. For large scale cloud based training, this can help in significantly reducing the amount of storage required along with lessening the computation and time for each retraining session.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

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

Citations in the corpus

Imbalanced Continual Learning with Partitioning Reservoir Sampling (2020, methodology): “We test (a) replay-based reservoir sampling [7,73,56] and (b) EWC [43] with a shared output head and a memory size of 1000.”
Compacting, Picking and Growing for Unforgetting Continual Learning (2019, methodology): “Data-preserving approaches (such as [32, 3, 11, 34]) are designed to directly save important data or latent codes as an efficient form, while memory-replay approaches [41, 46, 13, 45, 27] introduce additional memory models such as GANs for keeping data information or distribution in an indirect way.”
Variational Prototype Replays for Continual Learning (2019, background): “There have been some attempts to select representative image examples to store based on different scoring functions (Chen et al., 2012; Koh & Liang, 2017; Brahma & Othon, 2018).”
Few-Shot Self Reminder to Overcome Catastrophic Forgetting (2018, background): “There has been some attempt at designing submodular scoring functions for subset selection in continual learning [28].”
Prototype Reminding for Continual Learning (2019, background): “There have been some attempts [7, 19, 5] selecting representative examples for D̃t based on different scoring functions.”