Our paper titled “A Wholistic View of Continual Learning with Deep Neural Networks: Forgotten Lessons and the Bridge to Active and Open World Learning” is now published in Neural Networks, volume 160, Pages 306-336.
This paper is a comprehensive 30 page resource to both survey & start unifying views for new and old continual learners alike. It starts with a comprehensive survey of continual learning, active learning and open set recognition. It then outlines synergies and native interfaces beyond the typical catastrophic forgetting. To facilitate understanding, we outline a concrete instantiation of a neural network based model to combine the paradigms. Through a set of teaching examples, we then illustrate how this helps with avoiding forgetting, while being robust to unseen unknown data inputs in inference, being able to pick new data points to train on, handle corrupted & perturbed examples, and even optimize task curricula (if this is permitted/accessible by the available data input). We then use these insights to extend the existing definitions of continual learning, so that future work can strive to be more comprehensive.
For more information, read the full paper or see the abstract below:
Current deep learning methods are regarded as favorable if they empirically perform well on dedicated test sets. This mentality is seamlessly reflected in the resurfacing area of continual learning, where consecutively arriving data is investigated. The core challenge is framed as protecting previously acquired representations from being catastrophically forgotten. However, comparison of individual methods is nevertheless performed in isolation from the real world by monitoring accumulated benchmark test set performance. The closed world assumption remains predominant, i.e. models are evaluated on data that is guaranteed to originate from the same distribution as used for training. This poses a massive challenge as neural networks are well known to provide overconfident false predictions on unknown and corrupted instances. In this work we critically survey the literature and argue that notable lessons from open set recognition, identifying unknown examples outside of the observed set, and the adjacent field of active learning, querying data to maximize the expected performance gain, are frequently overlooked in the deep learning era. Hence, we propose a consolidated view to bridge continual learning, active learning and open set recognition in deep neural networks. Finally, the established synergies are supported empirically, showing joint improvement in alleviating catastrophic forgetting, querying data, selecting task orders, while exhibiting robust open world application.
Abstract of our paper “A Wholistic View of Continual Learning with Deep Neural Networks: Forgotten Lessons and the Bridge to Active and Open World Learning”