Continual Learning in Neural Networks

By Rahaf Aljundi. 2019

49 citations recorded by the source corpus.

Artificial neural networks have exceeded human-level performance in accomplishing several individual tasks (e.g. voice recognition, object recognition, and video games). However, such success remains modest compared to human intelligence that can learn and perform an unlimited number of tasks. Humans' ability of learning and accumulating knowledge over their lifetime is an essential aspect of their intelligence. Continual machine learning aims at a higher level of machine intelligence through providing the artificial agents with the ability to learn online from a non-stationary and never-ending stream of data. A key component of such a never-ending learning process is to overcome the catastrophic forgetting of previously seen data, a problem that neural networks are well known to suffer from. The work described in this thesis has been dedicated to the investigation of continual learning and solutions to mitigate the forgetting phenomena in neural networks. To approach the continual learning problem, we first assume a task incremental setting where tasks are received one at a time and data from previous tasks are not stored. Since the task incremental setting can't be assumed in all continual learning scenarios, we also study the more general online continual setting. We consider an infinite stream of data drawn from a non-stationary distribution with a supervisory or self-supervisory training signal. The proposed methods in this thesis have tackled important aspects of continual learning. They were evaluated on different benchmarks and over various learning sequences. Advances in the state of the art of continual learning have been shown and challenges for bringing continual learning into application were critically identified.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Continual Learning on Noisy Data Streams via Self-Purified Replay (2021, methodology): “We explicitly ground our experiment setting based on the recent suggestions for robust evaluation in continual learing [1, 18, 79] as follows.”
Imbalanced Continual Learning with Partitioning Reservoir Sampling (2020, background): “As we are the first to explore multi-label continual learning, we explicitly ground our experimental setting based on [26,1,81] as follows:”
Prototype Augmentation and Self-Supervision for Incremental Learning (2021, background): “At step t, the goal is to minimize a predefined loss function L on new dataset Dt without interfering with and possibly improving on those that were learned previously [1]:”
Pretrained Language Model in Continual Learning: A Comparative Study (2022, methodology): “Conversely, in class-incremental learning, the methods have to perform task inference, implicitly (Aljundi, 2019) or explicitly (Lee et al., 2020b).”
Continual Learning in Automatic Speech Recognition (2020, background): “Recent works on Continual Learning (CL) [10, 11, 12] aim at emulating the gradual learning observed in humans using machines by utilizing a continuous stream of data over a period of time to learn and share knowledge over different data domains.”
VariGrow: Variational Architecture Growing for Task-Agnostic Continual Learning based on Bayesian Novelty (2022, methodology): “This new setting is challenging, dubbed as the single-headed setting, where existing task-agnostic CL methods have significantly lower performance compared to their task-aware counterparts (Rao et al., 2019; Aljundi, 2019; Aljundi et al., 2019; Zeno et al., 2018; He & Jaeger, 2018).”