mlsanityData to 7 Oct 2026

Active Continual Learning with Metaplastic Binary Bayesian Neural Networks

arXiv 2605.30198 ↗First posted 28 May 2026cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Always-on edge systems must keep learning as conditions change under tight compute budgets and must detect unreliable predictions. Bayesian binary neural networks are attractive in this setting, but mean-field Bernoulli posteriors can saturate on long non-stationary streams, wiping out epistemic uncertainty and freezing plasticity. We propose BiMU, derived from a bounded-memory variational objective that balances stability, plasticity, and forgetting. BiMU combines a data term with controlled relaxation toward the prior and an uncertainty-dependent step size that prevents saturation and sustains informative uncertainty. This non-degenerate posterior enables fully online, buffer-free active querying via Monte Carlo disagreement, reducing label queries and backpropagation updates under imbalance. BiMU sustains learning and strong OOD detection on 1000-tasks Permuted-MNIST, and on OpenLORIS-Object achieves up to 32× label/update savings at matched accuracy under class imbalance and feature compression.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 28 of its 43 references that are in the map. First posted on arXiv 28 May 2026.

Builds on

28 of 43 references.

References · 28

  1. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  2. 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
  3. 2017Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithmscited by 4,283
  4. 2019Optuna: A Next-generation Hyperparameter Optimization Frameworkcited by 2,455
  5. 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,335
  6. 2017Gradient Episodic Memory for Continual Learningcited by 1,871
  7. 2016Instance Normalization: The Missing Ingredient for Fast Stylizationcited by 1,187
  8. 2016The Concrete Distribution: A Continuous Relaxation of Discrete Random Variablescited by 1,135
  9. 2017Deep Bayesian Active Learning with Image Datacited by 812
  10. 2018Progress & Compress: A scalable framework for continual learningcited by 516
  11. 2015Variational Dropout and the Local Reparameterization Trickcited by 489
  12. 2017Variational Continual Learningcited by 335
16 more
  1. 2013Revisiting Natural Gradient for Deep Networkscited by 157
  2. 2021The Bayesian Learning Rulecited by 95
  3. 2021Gone Fishing: Neural Active Learning with Fisher Embeddingscited by 88
  4. 2019Striking the Right Balance with Uncertaintycited by 79
  5. 2013Streaming Variational Bayescited by 46
  6. 2020Task Agnostic Continual Learning Using Online Variational Bayes with Fixed-Point Updatescited by 34
  7. 2018XNOR Neural Engine: a Hardware Accelerator IP for 21.6 fJ/op Binary Neural Network Inferencecited by 28
  8. 2019OpenLORIS-Object: A Robotic Vision Dataset and Benchmark for Lifelong Deep Learningcited by 28
  9. 2021Synaptic metaplasticity in binarized neural networkscited by 28
  10. 2020Training Binary Neural Networks using the Bayesian Learning Rulecited by 22
  11. 2021Sound Event Detection with Binary Neural Networks on Tightly Power-Constrained IoT Devicescited by 10
  12. 2023Elephant Neural Networks: Born to Be a Continual Learnercited by 10
  13. 2021Continual Active Learning for Efficient Adaptation of Machine Learning Models to Changing Image Acquisitioncited by 9
  14. 2025Bayesian continual learning and forgetting in neural networkscited by 8
  15. 2025Active Learning for Continual Learning: Keeping the Past Alive in the Presentcited by 5
  16. 2023Active Continual Learning: On Balancing Knowledge Retention and Learnabilitycited by 3

Built on by

Not cited yet.