# Generative replay with feedback connections as a general strategy for continual learning

arXiv: https://arxiv.org/abs/1809.10635 · HTML: https://mlsanity.com/paper/1809.10635/

## Key facts

- Cited by 104 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 27 Sep 2018.

## Built on by (most cited first, 30 of 104)

- [CODA-Prompt: COntinual Decomposed Attention-based Prompting for Rehearsal-Free Continual Learning](https://mlsanity.com/paper/2211.13218/) (2022)
- [Online Continual Learning in Image Classification: An Empirical Survey](https://mlsanity.com/paper/2101.10423/) (2021)
- [Always Be Dreaming: A New Approach for Data-Free Class-Incremental Learning](https://mlsanity.com/paper/2106.09701/) (2021)
- [Avalanche: an End-to-End Library for Continual Learning](https://mlsanity.com/paper/2104.00405/) (2021)
- [Wild-Time: A Benchmark of in-the-Wild Distribution Shift over Time](https://mlsanity.com/paper/2211.14238/) (2022)
- [Recent Advances of Continual Learning in Computer Vision: An Overview](https://mlsanity.com/paper/2109.11369/) (2021)
- [A Closer Look at Rehearsal-Free Continual Learning](https://mlsanity.com/paper/2203.17269/) (2022)
- [Powerpropagation: A sparsity inducing weight reparameterisation](https://mlsanity.com/paper/2110.00296/) (2021)
- [Natural continual learning: success is a journey, not (just) a destination](https://mlsanity.com/paper/2106.08085/) (2021)
- [Posterior Meta-Replay for Continual Learning](https://mlsanity.com/paper/2103.01133/) (2021)
- [A Data-Free Approach to Mitigate Catastrophic Forgetting in Federated Class Incremental Learning for Vision Tasks](https://mlsanity.com/paper/2311.07784/) (2023)
- [Class-Incremental Learning with Generative Classifiers](https://mlsanity.com/paper/2104.10093/) (2021)
- [DualNet: Continual Learning, Fast and Slow](https://mlsanity.com/paper/2110.00175/) (2021)
- [Distilled Replay: Overcoming Forgetting through Synthetic Samples](https://mlsanity.com/paper/2103.15851/) (2021)
- [Non-IID data and Continual Learning processes in Federated Learning: A long road ahead](https://mlsanity.com/paper/2111.13394/) (2021)
- [Continual Learning of Knowledge Graph Embeddings](https://mlsanity.com/paper/2101.05850/) (2021)
- [Symbolic Replay: Scene Graph as Prompt for Continual Learning on VQA Task](https://mlsanity.com/paper/2208.12037/) (2022)
- [Continual Learning for Recurrent Neural Networks: an Empirical Evaluation](https://mlsanity.com/paper/2103.07492/) (2021)
- [Continual Learning via Bit-Level Information Preserving](https://mlsanity.com/paper/2105.04444/) (2021)
- [An Investigation of Replay-based Approaches for Continual Learning](https://mlsanity.com/paper/2108.06758/) (2021)
- [Hybrid Predictive Coding: Inferring, Fast and Slow](https://mlsanity.com/paper/2204.02169/) (2022)
- [Formalizing the Generalization-Forgetting Trade-off in Continual Learning](https://mlsanity.com/paper/2109.14035/) (2021)
- [Diffusion-Driven Data Replay: A Novel Approach to Combat Forgetting in Federated Class Continual Learning](https://mlsanity.com/paper/2409.01128/) (2024)
- [Unleashing the Power of Continual Learning on Non-Centralized Devices: A Survey](https://mlsanity.com/paper/2412.13840/) (2024)
- [CLeaR: An Adaptive Continual Learning Framework for Regression Tasks](https://mlsanity.com/paper/2101.00926/) (2021)
- [Condensed Composite Memory Continual Learning](https://mlsanity.com/paper/2102.09890/) (2021)
- [Continual Learning via Inter-Task Synaptic Mapping](https://mlsanity.com/paper/2106.13954/) (2021)
- [BioSLAM: A Bio-inspired Lifelong Memory System for General Place Recognition](https://mlsanity.com/paper/2208.14543/) (2022)
- [A Domain-Agnostic Approach for Characterization of Lifelong Learning Systems](https://mlsanity.com/paper/2301.07799/) (2023)
- [Do Not Forget to Attend to Uncertainty while Mitigating Catastrophic Forgetting](https://mlsanity.com/paper/2102.01906/) (2021)

## Abstract (arXiv)

A major obstacle to developing artificial intelligence applications capable of true lifelong learning is that artificial neural networks quickly or catastrophically forget previously learned tasks when trained on a new one. Numerous methods for alleviating catastrophic forgetting are currently being proposed, but differences in evaluation protocols make it difficult to directly compare their performance. To enable more meaningful comparisons, here we identified three distinct scenarios for continual learning based on whether task identity is known and, if it is not, whether it needs to be inferred. Performing the split and permuted MNIST task protocols according to each of these scenarios, we found that regularization-based approaches (e.g., elastic weight consolidation) failed when task identity needed to be inferred. In contrast, generative replay combined with distillation (i.e., using class probabilities as "soft targets") achieved superior performance in all three scenarios. Addressing the issue of efficiency, we reduced the computational cost of generative replay by integrating the generative model into the main model by equipping it with generative feedback or backward connections. This Replay-through-Feedback approach substantially shortened training time with no or negligible loss in performance. We believe this to be an important first step towards making the powerful technique of generative replay scalable to real-world continual learning applications.