Generative Continual Concept Learning

By Mohammad Rostami, Soheil Kolouri, James L. McClelland, Praveen K. Pilly. 2019

52 citations recorded by the source corpus.

After learning a concept, humans are also able to continually generalize their learned concepts to new domains by observing only a few labeled instances without any interference with the past learned knowledge. In contrast, learning concepts efficiently in a continual learning setting remains an open challenge for current Artificial Intelligence algorithms as persistent model retraining is necessary. Inspired by the Parallel Distributed Processing learning and the Complementary Learning Systems theories, we develop a computational model that is able to expand its previously learned concepts efficiently to new domains using a few labeled samples. We couple the new form of a concept to its past learned forms in an embedding space for effective continual learning. Doing so, a generative distribution is learned such that it is shared across the tasks in the embedding space and models the abstract concepts. This procedure enables the model to generate pseudo-data points to replay the past experience to tackle catastrophic forgetting.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Generative Models, Meta & Continual Learning

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

In Generative Models

In Meta & Continual Learning

Citations in the corpus

A Comprehensive Survey of Continual Learning: Theory, Method and Application (2023, background): “is often represented by a memory buffer [41], [364], [366], [431], [440] or a generative model [365], [383], [412], [427],”
Detection and Continual Learning of Novel Face Presentation Attacks (2021, methodology): “The idea is to enable the model to generate pseudo-data points that are similar to the data points of the past tasks and use the pseudo-data points for pseudo-rehearsal [33, 28, 27].”
Knowledge Capture and Replay for Continual Learning (2020, background): “On the other hand, generative replay approaches [1, 12, 22, 20] face challenges in scalability to complex datasets with many tasks, since they involve heavy training and computation associated with an auxiliary network to generate visually meaningful images for each task.”
Unsupervised Model Adaptation for Continual Semantic Segmentation (2020, background): “Note that this situation emerges naturally in continual learning (Shin et al. 2017; Rostami et al. 2020) when data distribution changes over time.”
Lifelong Infinite Mixture Model Based on Knowledge-Driven Dirichlet Process (2021, background): “The Generative Replay Mechanism (GRM) [41] is a popular lifelong learning approach showing promising results overcoming catastrophic forgetting [1, 33, 39, 48, 56].”
Is Multi-Task Learning an Upper Bound for Continual Learning? (2022, methodology): “…1) regularization-based approaches that penalize large changes to important parameters of previous tasks [1, 2, 3, 4, 5], and 2) memory replay and rehearsal-based approaches [6, 7, 8, 9], and 3) architectural methods that rely on model expansion, parameter isolation, and masking [10, 11, 12, 13].”
What shapes feature representations? Exploring datasets, architectures, and training (2020, background): “However, catastrophic interference between new tasks and prior tasks shows that at least some features are suppressed or altered during training [28, 21, 46, 33, 27].”
Using Task Descriptions in Lifelong Machine Learning for Improved Performance and Zero-Shot Transfer (2017, background): “…often through means of learning an underlying representation that extends to new tasks and uses outside information that appropriately maps to the latent space (Palatucci, Hinton, Pomerleau, & Mitchell, 2009; Socher, Ganjoo, Manning, & Ng, 2013; Rostami, Kolouri, McClelland, & Pilly, 2020).”
Sequential Unsupervised Domain Adaptation through Prototypical Distributions (2020, methodology): “This training procedure have been implemented using either adversarial learning [5–9] or by directly minimizing the distance between the two distributions in the embedding [10–13 , 13– 15].”
Continual Contrastive Learning for Cross-Dataset Scene Classification (2022, methodology): “[12] used a generative model to produce pseudo samples with a few samples from the previous tasks, so that the abstract concepts have effectively emigrated from the generative model to the current task.”
Learning Transferable Knowledge Through Embedding Spaces (2019, unknown): “Results of this chapter have been presented in [217; 218].”
FFNB: Forgetting-Free Neural Blocks for Deep Continual Learning (2021, methodology): “,T10 correspond to classes [51-100] while T0 is the pretraining task involving classes [1-50]).”