Pith. sign in

REVIEW 8 cited by

Maintaining Plasticity in Deep Continual Learning

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2306.13812 v3 pith:OQ6X4FRN submitted 2023-06-23 cs.LG

classification cs.LG
keywords plasticitycontinuallearninglossbackpropagationdeepdeep-learningexamples
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Modern deep-learning systems are specialized to problem settings in which training occurs once and then never again, as opposed to continual-learning settings in which training occurs continually. If deep-learning systems are applied in a continual learning setting, then it is well known that they may fail to remember earlier examples. More fundamental, but less well known, is that they may also lose their ability to learn on new examples, a phenomenon called loss of plasticity. We provide direct demonstrations of loss of plasticity using the MNIST and ImageNet datasets repurposed for continual learning as sequences of tasks. In ImageNet, binary classification performance dropped from 89% accuracy on an early task down to 77%, about the level of a linear network, on the 2000th task. Loss of plasticity occurred with a wide range of deep network architectures, optimizers, activation functions, batch normalization, dropout, but was substantially eased by L2-regularization, particularly when combined with weight perturbation. Further, we introduce a new algorithm -- continual backpropagation -- which slightly modifies conventional backpropagation to reinitialize a small fraction of less-used units after each example and appears to maintain plasticity indefinitely.

Discussion (0). Sign in to comment.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Predicting Plasticity in Deep Continual Learning: A Theoretical Perspective

    cs.LG 2026-05 unverdicted novelty 7.0 of 10

    Optimization readiness, defined from gradient strength and reliability, lower-bounds one-step optimization gain and outperforms rank-based diagnostics in predicting neural network trainability across continual learnin...

  2. When is Warmstarting Effective for Scaling Language Models?

    cs.LG 2026-05 unverdicted novelty 6.0 of 10

    A 2x growth factor in model warmstarting yields reliable training speedups for language models under 20 tokens/parameter budgets, with an empirical upper bound on effective growth factors.

  3. Adapt or Forget: Provable Tradeoffs Between Adam and SGD in Nonstationary Optimization

    stat.ML 2026-05 unverdicted novelty 6.0 of 10

    Adam's adaptive preconditioning and first-moment averaging improve high-probability tracking error in noise-dominated nonstationary regimes but can increase it under strong drift, where SGD achieves a smaller floor, w...

  4. FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control

    cs.LG 2026-04 unverdicted novelty 6.0 of 10

    FlashSAC scales up Soft Actor-Critic with fewer updates, larger models, higher data throughput, and norm bounds to deliver faster, more stable training than PPO on high-dimensional robot control tasks across dozens of...

  5. FlashSAC: Fast and Stable Off-Policy Reinforcement Learning for High-Dimensional Robot Control

    cs.LG 2026-04 unverdicted novelty 6.0 of 10

    FlashSAC improves training speed and final performance of off-policy RL on high-dimensional robot tasks by reducing update frequency, increasing model scale, and bounding norms to limit critic error accumulation.

  6. Attribution-Based Neuron Utility for Plasticity Restoration in Deep Networks

    cs.LG 2026-05 unverdicted novelty 5.0 of 10

    GXD estimates the first-order functional cost of replacing a neuron via gradient attribution to make adaptive resets more reliable for preserving plasticity in continual learning.

  7. RIZZ: Routing Interactions to Near Zero-Interference Zones for Continual Adaptation of Black-Box Agents

    cs.AI 2026-06 unverdicted novelty 4.0 of 10

    RIZZ is a continual adaptation framework for black-box LLM agents that uses dynamically spawned memory branches, context-aware routing, verifier-gated updates, and prompt compilation to control interference across non...

  8. Plasticity Loss in Deep Reinforcement Learning: A Survey

    cs.AI 2024-11 unverdicted novelty 4.0 of 10

    Survey unifies the definition of plasticity loss in DRL, taxonomizes over 50 mitigations, identifies evaluation gaps, and finds general regularization often outperforms domain-specific methods.

Pith tools