pith. sign in

arxiv: 1810.12488 · v4 · pith:NBJPKMXAnew · submitted 2018-10-30 · 💻 cs.LG · cs.AI· cs.CV

Re-evaluating Continual Learning Scenarios: A Categorization and Case for Strong Baselines

classification 💻 cs.LG cs.AIcs.CV
keywords scenariosbaselinescategorizationcontinuallearningmethodsseveralstrong
0
0 comments X
read the original abstract

Continual learning has received a great deal of attention recently with several approaches being proposed. However, evaluations involve a diverse set of scenarios making meaningful comparison difficult. This work provides a systematic categorization of the scenarios and evaluates them within a consistent framework including strong baselines and state-of-the-art methods. The results provide an understanding of the relative difficulty of the scenarios and that simple baselines (Adagrad, L2 regularization, and naive rehearsal strategies) can surprisingly achieve similar performance to current mainstream methods. We conclude with several suggestions for creating harder evaluation scenarios and future research directions. The code is available at https://github.com/GT-RIPL/Continual-Learning-Benchmark

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 11 Pith papers

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

  1. Optimal L2 Regularization in High-dimensional Continual Linear Regression

    cs.LG 2026-01 unverdicted novelty 8.0

    In high-dimensional continual linear regression, optimal fixed L2 regularization strength scales as T/ln T with the number of tasks and mitigates label noise for arbitrary linear teachers.

  2. Stop Marginalizing My Dreams: Model Inversion via Laplace Kernel for Continual Learning

    cs.LG 2026-05 unverdicted novelty 7.0

    REMIX uses Laplace kernel parameterization to enable scalable full-covariance modeling in model inversion, improving synthetic sample quality and performance in data-free continual learning.

  3. Exemplar-Free Continual Learning for State Space Models

    cs.LG 2025-05 unverdicted novelty 7.0

    Inf-SSM constrains the infinite-horizon evolution of SSMs via Grassmannian geometry and an efficient O(n^2) Sylvester solver to enable exemplar-free continual learning with reduced forgetting.

  4. Temporal Taskification in Streaming Continual Learning: A Source of Evaluation Instability

    cs.LG 2026-04 conditional novelty 6.0

    Different valid temporal partitions of the same streaming dataset can produce materially different rankings and performance numbers for continual learning methods.

  5. Fine-Tuning Regimes Define Distinct Continual Learning Problems

    cs.LG 2026-04 unverdicted novelty 6.0

    The relative rankings of continual learning methods are not preserved across different fine-tuning regimes defined by trainable parameter depth.

  6. Continual Distillation of Teachers from Different Domains

    cs.LG 2026-04 conditional novelty 6.0

    SE2D stabilizes continual distillation across heterogeneous teachers by preserving logits on external unlabeled data to mitigate unseen knowledge forgetting.

  7. Tracking Adaptation Time: Metrics for Temporal Distribution Shift

    cs.LG 2026-04 unverdicted novelty 6.0

    Three complementary metrics are introduced to distinguish model adaptation from intrinsic data difficulty under temporal distribution shift.

  8. Robust Policy Optimization to Prevent Catastrophic Forgetting

    cs.LG 2026-02 unverdicted novelty 6.0

    FRPO applies a max-min robust optimization over KL-bounded policy neighborhoods during RLHF to reduce catastrophic forgetting of safety and accuracy under subsequent SFT or RL fine-tuning.

  9. A Survey of Continual Reinforcement Learning

    cs.LG 2025-06 accept novelty 6.0

    The paper surveys CRL literature, proposes a taxonomy of methods into four categories based on knowledge storage and transfer, reviews metrics and benchmarks, and outlines challenges and future research directions.

  10. Autoencoder-Based Incremental Class Learning without Retraining on Old Data

    cs.LG 2019-07 unverdicted novelty 4.0

    Autoencoder extracts class prototypes whose means enable metric classification in incremental learning, matching SOTA accuracy with lower memory overhead on CIFAR-100 and CUB-200-2011 via regularization to avoid forgetting.

  11. Neural Network Optimization Reimagined: Decoupled Techniques for Scratch and Fine-Tuning

    cs.CV 2026-04 unverdicted novelty 3.0

    DualOpt decouples optimization by using real-time layer-wise weight decay for scratch training and weight rollback for fine-tuning to improve convergence, generalization, and reduce knowledge forgetting.