Generalized Variational Continual Learning

Continual learning deals with training models on new tasks and datasets in an online fashion. One strand of research has used probabilistic regularization for continual learning, with two of the main approaches in this vein being Online Elastic Weight Consolidation (Online EWC) and Variational Continual Learning (VCL). VCL employs variational inference, which in other settings has been improved empirically by applying likelihood-tempering. We show that applying this modification to VCL recovers Online EWC as a limiting case, allowing for interpolation between the two approaches. We term the general algorithm Generalized VCL (GVCL). In order to mitigate the observed overpruning effect of VI, we take inspiration from a common multi-task architecture, neural networks with task-specific FiLM layers, and find that this addition leads to significant performance gains, specifically for variational methods. In the small-data regime, GVCL strongly outperforms existing baselines. In larger datasets, GVCL with FiLM layers outperforms or is competitive with existing baselines in terms of accuracy, whilst also providing significantly better calibration.

[1]  James Martens,et al.  New Insights and Perspectives on the Natural Gradient Method , 2014, J. Mach. Learn. Res..

[2]  Sebastian Nowozin,et al.  How Good is the Bayes Posterior in Deep Neural Networks Really? , 2020, ICML.

[3]  Aaron C. Courville,et al.  FiLM: Visual Reasoning with a General Conditioning Layer , 2017, AAAI.

[4]  R. French.  Catastrophic forgetting in connectionist networks , 1999, Trends in Cognitive Sciences.

[5]  Richard E. Turner,et al.  Overpruning in Variational Bayesian Neural Networks , 2018, 1801.06230.

[6]  Manfred Opper,et al.  The Variational Gaussian Approximation Revisited , 2009, Neural Computation.

[7]  Taesup Moon,et al.  Uncertainty-based Continual Learning with Adaptive Regularization , 2019, NeurIPS.

[8]  David Barber,et al.  Online Structured Laplace Approximations For Overcoming Catastrophic Forgetting , 2018, NeurIPS.

[9]  Dmitry P. Vetrov,et al.  Variational Dropout Sparsifies Deep Neural Networks , 2017, ICML.

[10]  Alexander A. Alemi,et al.  Fixing a Broken ELBO , 2017, ICML.

[11]  Razvan Pascanu,et al.  Progressive Neural Networks , 2016, ArXiv.

[12]  Razvan Pascanu,et al.  Overcoming catastrophic forgetting in neural networks , 2016, Proceedings of the National Academy of Sciences.

[13]  Marc'Aurelio Ranzato,et al.  Gradient Episodic Memory for Continual Learning , 2017, NIPS.

[14]  Yee Whye Teh,et al.  Progress & Compress: A scalable framework for continual learning , 2018, ICML.

[15]  Richard E. Turner,et al.  Two problems with variational expectation maximisation for time-series models , 2011 .

[16]  Christopher Burgess,et al.  beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework , 2016, ICLR 2016.

[17]  Mohammad Emtiyaz Khan,et al.  Continual Deep Learning by Functional Regularisation of Memorable Past , 2020, NeurIPS.

[18]  Byoung-Tak Zhang,et al.  Overcoming Catastrophic Forgetting by Incremental Moment Matching , 2017, NIPS.

[19]  Guodong Zhang,et al.  Noisy Natural Gradient as Variational Inference , 2017, ICML.

[20]  Sebastian Nowozin,et al.  Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes , 2019, NeurIPS.

[21]  Mohammad Emtiyaz Khan,et al.  Practical Deep Learning with Bayesian Principles , 2019, NeurIPS.

[22]  Alexander J. Smola,et al.  Laplace Propagation , 2003, NIPS.

[23]  Chrisantha Fernando,et al.  PathNet: Evolution Channels Gradient Descent in Super Neural Networks , 2017, ArXiv.

[24]  Mehrdad Farajtabar,et al.  SOLA: Continual Learning with Second-Order Loss Approximation , 2020, ArXiv.

[25]  Richard E. Turner,et al.  Improving and Understanding Variational Continual Learning , 2019, ArXiv.

[26]  Surya Ganguli,et al.  Continual Learning Through Synaptic Intelligence , 2017, ICML.

[27]  Richard E. Turner,et al.  Continual Learning with Adaptive Weights (CLAW) , 2020, ICLR.

[28]  Richard E. Turner,et al.  Variational Continual Learning , 2017, ICLR.

[29]  Andrea Vedaldi,et al.  Learning multiple visual domains with residual adapters , 2017, NIPS.

[30]  Stefano Soatto,et al.  Where is the Information in a Deep Neural Network? , 2019, ArXiv.

[31]  Alexandros Karatzoglou,et al.  Overcoming Catastrophic Forgetting with Hard Attention to the Task , 2018 .