English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Continual learning aims to enable a model to incrementally learn knowledgefrom sequentially arrived data. Previous works adopt the conventionalclassification architecture, which consists of a feature extractor and aclassifier. The feature extractor is shared across sequentially arrived tasksor classes, but one specific group of weights of the classifier correspondingto one new class should be incrementally expanded. Consequently, the parametersof a continual learner gradually increase. Moreover, as the classifier containsall historical arrived classes, a certain size of the memory is usuallyrequired to store rehearsal data to mitigate classifier bias and catastrophicforgetting. In this paper, we propose a non-incremental learner, namedAttriCLIP, to incrementally extract knowledge of new classes or tasks.Specifically, AttriCLIP is built upon the pre-trained visual-language modelCLIP. Its image encoder and text encoder are fixed to extract features fromboth images and text. Text consists of a category name and a fixed number oflearnable parameters which are selected from our designed attribute word bankand serve as attributes. As we compute the visual and textual similarity forclassification, AttriCLIP is a non-incremental learner. The attribute prompts,which encode the common knowledge useful for classification, can effectivelymitigate the catastrophic forgetting and avoid constructing a replay memory. Weevaluate our AttriCLIP and compare it with CLIP-based and previousstate-of-the-art continual learning methods in realistic settings withdomain-shift and long-sequence learning. The results show that our methodperforms favorably against previous state-of-the-arts. The implementation codecan be available at https://github.com/bhrqw/AttriCLIP.

AttriCLIP: A Non-Incremental Learner for Incremental Knowledge Learning