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

Recently, Vision Language Models (VLMs) have gained significant attention,exhibiting notable advancements across various tasks by leveraging extensiveimage-text paired data. However, prevailing VLMs often treat Visual QuestionAnswering (VQA) as perception tasks, employing black-box models that overlookexplicit modeling of relationships between different questions within the samevisual scene. Moreover, the existing VQA methods that rely on Knowledge Bases(KBs) might frequently encounter biases from limited data and face challengesin relevant information indexing. Attempt to overcome these limitations, thispaper introduces an explainable multi-agent collaboration framework by tappinginto knowledge embedded in Large Language Models (LLMs) trained on extensivecorpora. Inspired by human cognition, our framework uncovers latent informationwithin the given question by employing three agents, i.e., Seeker, Responder,and Integrator, to perform a top-down reasoning process. The Seeker agentgenerates relevant issues related to the original question. The Responderagent, based on VLM, handles simple VQA tasks and provides candidate answers.The Integrator agent combines information from the Seeker agent and theResponder agent to produce the final VQA answer. Through the abovecollaboration mechanism, our framework explicitly constructs a multi-viewknowledge base for a specific image scene, reasoning answers in a top-downprocessing manner. We extensively evaluate our method on diverse VQA datasetsand VLMs, demonstrating its broad applicability and interpretability withcomprehensive experimental results.

Towards Top-Down Reasoning: An Explainable Multi-Agent Approach for Visual Question Answering