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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

VQA is an ambitious task aiming to answer any image-related question.However, in reality, it is hard to build such a system once for all since theneeds of users are continuously updated, and the system has to implement newfunctions. Thus, Continual Learning (CL) ability is a must in developingadvanced VQA systems. Recently, a pioneer work split a VQA dataset intodisjoint answer sets to study this topic. However, CL on VQA involves not onlythe expansion of label sets (new Answer sets). It is crucial to study how toanswer questions when deploying VQA systems to new environments (new Visualscenes) and how to answer questions requiring new functions (new Questiontypes). Thus, we propose CLOVE, a benchmark for Continual Learning On VisualquEstion answering, which contains scene- and function-incremental settings forthe two aforementioned CL scenarios. In terms of methodology, the maindifference between CL on VQA and classification is that the former additionallyinvolves expanding and preventing forgetting of reasoning mechanisms, while thelatter focusing on class representation. Thus, we propose a real-data-freereplay-based method tailored for CL on VQA, named Scene Graph as Prompt forSymbolic Replay. Using a piece of scene graph as a prompt, it replays pseudoscene graphs to represent the past images, along with correlated QA pairs. Aunified VQA model is also proposed to utilize the current and replayed data toenhance its QA ability. Finally, experimental results reveal challenges inCLOVE and demonstrate the effectiveness of our method. The dataset and codewill be available at https://github.com/showlab/CLVQA.

Symbolic Replay: Scene Graph as Prompt for Continual Learning on VQA Task