We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Uncertainty-based Continual Learning with Adaptive Regularization

Abstract: We introduce a new neural network-based continual learning algorithm, dubbed as Uncertainty-regularized Continual Learning (UCL), which builds on traditional Bayesian online learning framework with variational inference. We focus on two significant drawbacks of the recently proposed regularization-based methods: a) considerable additional memory cost for determining the per-weight regularization strengths and b) the absence of gracefully forgetting scheme, which can prevent performance degradation in learning new tasks. In this paper, we show UCL can solve these two problems by introducing a fresh interpretation on the Kullback-Leibler (KL) divergence term of the variational lower bound for Gaussian mean-field approximation. Based on the interpretation, we propose the notion of node-wise uncertainty, which drastically reduces the number of additional parameters for implementing per-weight regularization. Moreover, we devise two additional regularization terms that enforce stability by freezing important parameters for past tasks and allow plasticity by controlling the actively learning parameters for a new task. Through extensive experiments, we show UCL convincingly outperforms most of recent state-of-the-art baselines not only on popular supervised learning benchmarks, but also on challenging lifelong reinforcement learning tasks. The source code of our algorithm is available at this https URL
Comments: 10 pages (including Supplementary Materials), Neurips 2019 camera ready version
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:1905.11614 [cs.LG]
  (or arXiv:1905.11614v3 [cs.LG] for this version)

Submission history

From: Sungmin Cha [view email]
[v1] Tue, 28 May 2019 05:21:07 GMT (550kb,D)
[v2] Sun, 27 Oct 2019 20:19:47 GMT (1737kb,D)
[v3] Thu, 14 Nov 2019 05:25:59 GMT (1738kb,D)

Link back to: arXiv, form interface, contact.