Using Hindsight to Anchor Past Knowledge in Continual Learning

Chaudhry, Arslan; Gordo, Albert; Dokania, Puneet K.; Han, Junwei; López-Paz, David

doi:10.48550/arxiv.2002.08165

Cited by 17 publications

(28 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The fifth setting is on CIFAR10, where we use the commonly used setting from GMED [14], which divides CI-FAR10 into 5 tasks equally. And we compare with online continual learning methods: AGEM [5], BGD [36], GEM [22], GSS-Greedy [3], HAL [4], ER [31], MIR [1], and GMED [14].…”

Section: Methodsmentioning

confidence: 99%

ACAE-REMIND for Online Continual Learning with Compressed Feature Replay

Wang¹,

Herranz²,

Weijer³

2021

Preprint

View full text Add to dashboard Cite

Online continual learning aims to learn from a non-IID stream of data from a number of different tasks, where the learner is only allowed to consider data once. Methods are typically allowed to use a limited buffer to store some of the images in the stream. Recently, it was found that feature replay, where an intermediate layer representation of the image is stored (or generated) leads to superior results than image replay, while requiring less memory. Quantized exemplars can further reduce the memory usage. However, a drawback of these methods is that they use a fixed (or very intransigent) backbone network. This significantly limits the learning of representations that can discriminate between all tasks. To address this problem, we propose an auxiliary classifier auto-encoder (ACAE) module for feature replay at intermediate layers with high compression rates. The reduced memory footprint per image allows us to save more exemplars for replay. In our experiments, we conduct task-agnostic evaluation under online continual learning setting and get state-of-the-art performance on ImageNet-Subset, CIFAR100 and CIFAR10 dataset.

show abstract

Section: Methodsmentioning

confidence: 99%

ACAE-REMIND for Online Continual Learning with Compressed Feature Replay

Wang¹,

Herranz²,

Weijer³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Rehearsal-based methods [7,8,16] construct a data buffer to save samples from older tasks to train with data from the current task. Based on this simple yet effective idea, many recent methods improve upon it by involv-ing additional knowledge distillation penalties [3,6,46,57], or leveraging self-supervised learning techniques [4,42]. Albeit its simplicity in concept, rehearsal-based methods achieve state-of-the-art performance on various benchmarks [32,40].…”

Section: Related Workmentioning

confidence: 99%

Learning to Prompt for Continual Learning

Wang¹,

Zhang²,

Lee³

et al. 2021

Preprint

View full text Add to dashboard Cite

The mainstream paradigm behind continual learning has been to adapt the model parameters to non-stationary data distributions, where catastrophic forgetting is the central challenge. Typical methods rely on a rehearsal buffer or known task identity at test time to retrieve learned knowledge and address forgetting, while this work presents a new paradigm for continual learning that aims to train a more succinct memory system without accessing task identity at test time. Our method learns to dynamically prompt (L2P) a pre-trained model to learn tasks sequentially under different task transitions. In our proposed framework, prompts are small learnable parameters, which are maintained in a memory space. The objective is to optimize prompts to instruct the model prediction and explicitly manage task-invariant and task-specific knowledge while maintaining model plasticity. We conduct comprehensive experiments under popular image classification benchmarks with different challenging continual learning settings, where L2P consistently outperforms prior state-ofthe-art methods. Surprisingly, L2P achieves competitive results against rehearsal-based methods even without a rehearsal buffer and is directly applicable to challenging taskagnostic continual learning. Source code is available at https://github.com/google-research/l2p.

show abstract

“…Continual Learning Following Lange et al [2019], we review the related methods to alleviate catastrophic forgetting in continual learning in three different but overlapping categories. Replaybased methods store and rephrase a memory of the examples or knowledge learned so far [Rebuffi et al, 2016, Lopez-Paz and Ranzato, 2017, Shin et al, 2017, Riemer et al, 2018, Rios and Itti, 2018, Zhang et al, 2019, Chaudhry et al, 2020. Regularization-based methods constrain the parameter updates while learning new tasks to preserve previous knowledge.…”

Section: Related Workmentioning

confidence: 99%

Task-agnostic Continual Learning with Hybrid Probabilistic Models

Kirichenko,

Farajtabar,

Rao

et al. 2021

Preprint

View full text Add to dashboard Cite

Learning new tasks continuously without forgetting on a constantly changing data distribution is essential for real-world problems but extremely challenging for modern deep learning. In this work we propose HCL, a Hybrid generative-discriminative approach to Continual Learning for classification. We model the distribution of each task and each class with a normalizing flow. The flow is used to learn the data distribution, perform classification, identify task changes, and avoid forgetting, all leveraging the invertibility and exact likelihood which are uniquely enabled by the normalizing flow model. We use the generative capabilities of the flow to avoid catastrophic forgetting through generative replay and a novel functional regularization technique. For task identification, we use state-of-the-art anomaly detection techniques based on measuring the typicality of the model's statistics. We demonstrate the strong performance of HCL on a range of continual learning benchmarks such as split-MNIST, split-CIFAR, and SVHN-MNIST. * Work partially done as an intern at DeepMind.

show abstract

Using Hindsight to Anchor Past Knowledge in Continual Learning

Cited by 17 publications

References 0 publications

ACAE-REMIND for Online Continual Learning with Compressed Feature Replay

ACAE-REMIND for Online Continual Learning with Compressed Feature Replay

Learning to Prompt for Continual Learning

Task-agnostic Continual Learning with Hybrid Probabilistic Models

Contact Info

Product

Resources

About