# Continuous Learning in Single-Incremental-Task Scenarios

Source: [Continuous Learning in Single-Incremental-Task Scenarios](<https://www.sota2.com/research/paper/68fa654a1e209131cea09463>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-25T21:22:52.893Z

Authors: Davide Maltoni, Vincenzo Lomonaco

Published: 2018-06-22T09:22:42.000Z

## Abstract

It was recently shown that architectural, regularization and rehearsal strategies can be used to train deep models sequentially on a number of disjoint tasks without forgetting previously acquired knowledge. However, these strategies are still unsatisfactory if the tasks are not disjoint but constitute a single incremental task (e.g., class-incremental learning). In this paper we point out the differences between multi-task and single-incremental-task scenarios and show that well-known approaches such as LWF, EWC and SI are not ideal for incremental task scenarios. A new approach, denoted as AR1, combining architectural and regularization strategies is then specifically proposed. AR1 overhead (in term of memory and computation) is very small thus making it suitable for online learning. When tested on CORe50 and iCIFAR-100, AR1 outperformed existing regularization strategies by a good margin.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Instance Classification on CORe50 (instance)](<https://www.sota2.com/research/sota/instance-classification-on-core50-instance.md>): Instance Classification; CORe50 (instance); Accuracy: 69.48.

- [Category Classification on iCIFAR-100 (category)](<https://www.sota2.com/research/sota/category-classification-on-icifar-100-category.md>): Category Classification; iCIFAR-100 (category); Avg Accuracy: 42.39.

- [Category Classification on HOWS-CL-25 long category](<https://www.sota2.com/research/sota/category-classification-on-hows-cl-25-long-category.md>): Category Classification; HOWS-CL-25 long category; Average Accuracy: 8.41.

- [Category Classification on HOWS-CL-25 category](<https://www.sota2.com/research/sota/category-classification-on-hows-cl-25-category.md>): Category Classification; HOWS-CL-25 category; Average Accuracy: 8.59.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/1806.08568>)

[PDF](<https://arxiv.org/pdf/1806.08568.pdf>)