Bring Evanescent Representations to Life in Lifelong Class Incremental Learning

By Marco Toldo, Mete Ozay. 2022

51 citations recorded by the source corpus.

In Class Incremental Learning (CIL), a classification model is progressively trained at each incremental step on an evolving dataset of new classes, while at the same time, it is required to preserve knowledge of all the classes ob-served so far. Prototypical representations can be lever-aged to model feature distribution for the past data and in-ject information of former classes in later incremental steps without resorting to stored exemplars. However, if not up-dated, those representations become increasingly outdated as the incremental learning progresses with new classes. To address the aforementioned problems, we propose a frame-work which aims to (i) model the semantic drift by learning the relationship between representations of past and novel classes among incremental steps, and (ii) estimate the feature drift, defined as the evolution of the represen-tations learned by models at each incremental step. Se-mantic and feature drifts are then jointly exploited to infer up-to-date representations of past classes (evanescent rep-resentations), and thereby infuse past knowledge into incre-mental training. We experimentally evaluate our framework achieving exemplar-free SotA results on multiple bench-marks. In the ablation study, we investigate nontrivial relationships between evanescent representations and models.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning, 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 Representation & Self-Supervised Learning

In Meta & Continual Learning

Citations in the corpus

A Comprehensive Survey of Continual Learning: Theory, Method and Application (2023, methodology): “Other methods exploit the class-wise statistics of feature representations to obtain a balanced classiﬁer, such as by imposing the representations to be transferable and invariant [173], [176], or compensating explicitly the representation shifts [174], [249].”
Resurrecting Old Classes with New Data for Exemplar-Free Continual Learning (2024, unknown): “Recent methods [28, 42, 46] also explored how to update the prototypes learned in old tasks to counter the drift.”
Elastic Feature Consolidation for Cold Start Exemplar-free Incremental Learning (2024, methodology): “Specifically we consider EWC (Kirkpatrick et al., 2017), LwF (Li & Hoiem, 2017), PASS (Zhu et al., 2021b), Fusion (Toldo & Ozay, 2022), FeTrIL (Petit et al., 2023) and SSRE (Zhu et al., 2022).”