Revisiting class-incremental learning with pre-trained models: Generalizability and adaptivity are all you need

DW Zhou, ZW Cai, HJ Ye, DC Zhan, Z Liu - arxiv preprint arxiv …, 2023 - arxiv.org
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 …

Revisiting class-incremental learning with pre-trained models: Generalizability and adaptivity are all you need

DW Zhou, ZW Cai, HJ Ye, DC Zhan, Z Liu - International Journal of …, 2024 - Springer
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 …

Ranpac: Random projections and pre-trained models for continual learning

MD McDonnell, D Gong, A Parvaneh… - Advances in …, 2024 - proceedings.neurips.cc
Continual learning (CL) aims to incrementally learn different tasks (such as classification) in
a non-stationary data stream without forgetting old ones. Most CL works focus on tackling …

Few-shot class-incremental learning via training-free prototype calibration

QW Wang, DW Zhou, YK Zhang… - Advances in Neural …, 2024 - proceedings.neurips.cc
Real-world scenarios are usually accompanied by continuously appearing classes with
scare labeled samples, which require the machine learning model to incrementally learn …

Generative Multi-modal Models are Good Class Incremental Learners

X Cao, H Lu, L Huang, X Liu… - Proceedings of the …, 2024 - openaccess.thecvf.com
In class incremental learning (CIL) scenarios the phenomenon of catastrophic forgetting
caused by the classifier's bias towards the current task has long posed a significant …

Expandable subspace ensemble for pre-trained model-based class-incremental learning

DW Zhou, HL Sun, HJ Ye… - Proceedings of the IEEE …, 2024 - openaccess.thecvf.com
Abstract Class-Incremental Learning (CIL) requires a learning system to continually learn
new classes without forgetting. Despite the strong performance of Pre-Trained Models …

Few-shot class incremental learning with attention-aware self-adaptive prompt

C Liu, Z Wang, T **ong, R Chen, Y Wu, J Guo… - … on Computer Vision, 2024 - Springer
Abstract Few-Shot Class-Incremental Learning (FSCIL) models aim to incrementally learn
new classes with scarce samples while preserving knowledge of old ones. Existing FSCIL …

Continual learning with pre-trained models: A survey

DW Zhou, HL Sun, J Ning, HJ Ye, DC Zhan - arxiv preprint arxiv …, 2024 - arxiv.org
Nowadays, real-world applications often face streaming data, which requires the learning
system to absorb new knowledge as data evolves. Continual Learning (CL) aims to achieve …

A Practitioner's Guide to Continual Multimodal Pretraining

K Roth, V Udandarao, S Dziadzio, A Prabhu… - arxiv preprint arxiv …, 2024 - arxiv.org
Multimodal foundation models serve numerous applications at the intersection of vision and
language. Still, despite being pretrained on extensive data, they become outdated over time …

Class-incremental learning with clip: Adaptive representation adjustment and parameter fusion

L Huang, X Cao, H Lu, X Liu - European Conference on Computer Vision, 2024 - Springer
Class-incremental learning is a challenging problem, where the goal is to train a model that
can classify data from an increasing number of classes over time. With the advancement of …