Continual Learning with Dual Regularizations

Continual learning (CL) has received a great amount of attention in recent years and a multitude of continual learning approaches arose. In this paper, we propose a continual learning approach with dual regularizations to alleviate the well-known issue of catastrophic forgetting in a challenging continual learning scenario – domain incremental learning. We reserve a buffer of past examples, dubbed memory set, to retain some information about previous tasks. The key idea is to regularize the learned representation space as well as the model outputs by utilizing the memory set based on interleaving the memory examples into the current training process. We verify our approach on four CL dataset benchmarks. Our experimental results demonstrate that the proposed approach is consistently superior to the compared methods on all benchmarks, especially in the case of small buffer size.

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

[2]  Marcus Rohrbach,et al.  Memory Aware Synapses: Learning what (not) to forget , 2017, ECCV.

[3]  Anthony V. Robins,et al.  Catastrophic Forgetting, Rehearsal and Pseudorehearsal , 1995, Connect. Sci..

[4]  Ioannis Mitliagkas,et al.  Manifold Mixup: Better Representations by Interpolating Hidden States , 2018, ICML.

[5]  Simone Calderara,et al.  Dark Experience for General Continual Learning: a Strong, Simple Baseline , 2020, NeurIPS.

[6]  Christoph H. Lampert,et al.  iCaRL: Incremental Classifier and Representation Learning , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[7]  Yoshua Bengio,et al.  An Empirical Investigation of Catastrophic Forgeting in Gradient-Based Neural Networks , 2013, ICLR.

[8]  Jiwon Kim,et al.  Continual Learning with Deep Generative Replay , 2017, NIPS.

[9]  Sebastian Thrun,et al.  A lifelong learning perspective for mobile robot control , 1994, Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS'94).

[10]  Hongyi Zhang,et al.  mixup: Beyond Empirical Risk Minimization , 2017, ICLR.

[11]  Geoffrey E. Hinton,et al.  Distilling the Knowledge in a Neural Network , 2015, ArXiv.

[12]  Gerald Tesauro,et al.  Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference , 2018, ICLR.

[13]  Andreas S. Tolias,et al.  Generative replay with feedback connections as a general strategy for continual learning , 2018, ArXiv.

[14]  Marc'Aurelio Ranzato,et al.  Continual Learning with Tiny Episodic Memories , 2019, ArXiv.

[15]  Derek Hoiem,et al.  Learning without Forgetting , 2016, IEEE Transactions on Pattern Analysis and Machine Intelligence.

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

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

[18]  Mark B. Ring.  Continual learning in reinforcement environments , 1995, GMD-Bericht.

[19]  Michael McCloskey,et al.  Catastrophic Interference in Connectionist Networks: The Sequential Learning Problem , 1989 .

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

[21]  Yen-Cheng Liu,et al.  Re-evaluating Continual Learning Scenarios: A Categorization and Case for Strong Baselines , 2018, ArXiv.

[22]  R Ratcliff,et al.  Connectionist models of recognition memory: constraints imposed by learning and forgetting functions. , 1990, Psychological review.

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

[24]  Philip H. S. Torr,et al.  Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigence , 2018, ECCV.

[25]  Marc'Aurelio Ranzato,et al.  Efficient Lifelong Learning with A-GEM , 2018, ICLR.

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

[27]  Andreas S. Tolias,et al.  Three scenarios for continual learning , 2019, ArXiv.