To cope with real-world dynamics, an intelligent system needs to incrementally acquire,
update, accumulate, and exploit knowledge throughout its lifetime. This ability, known as …

Abstract Machine learning methods strive to acquire a robust model during the training
process that can effectively generalize to test samples, even in the presence of distribution …

Incrementally learning new information from a non-stationary stream of data, referred to as
'continual learning', is a key feature of natural intelligence, but a challenging problem for …

Class-incremental learning (CIL) aims to adapt to emerging new classes without forgetting
old ones. Traditional CIL models are trained from scratch to continually acquire knowledge …

The landscape of publicly available vision foundation models (VFMs) such as CLIP and
SAM is expanding rapidly. VFMs are endowed with distinct capabilities stemming from their …

Test-time domain adaptation aims to adapt a source pre-trained model to a target domain
without using any source data. Existing works mainly consider the case where the target …

Continual learning aims to enable a single model to learn a sequence of tasks without
catastrophic forgetting. Top-performing methods usually require a rehearsal buffer to store …

The mainstream paradigm behind continual learning has been to adapt the model
parameters to non-stationary data distributions, where catastrophic forgetting is the central …

Large pre-trained models such as CLIP or ALIGN offer consistent accuracy across a range of
data distributions when performing zero-shot inference (ie, without fine-tuning on a specific …

Biological organisms learn from interactions with their environment throughout their lifetime.
For artificial systems to successfully act and adapt in the real world, it is desirable to similarly …