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

To encourage AI agents to conduct meaningful Visual Dialogue (VD), the use ofReinforcement Learning has been proven potential. In Reinforcement Learning, itis crucial to represent states and assign rewards based on the action-causedtransitions of states. However, the state representation in previous VisualDialogue works uses the textual information only and its transitions areimplicit. In this paper, we propose Explicit Concerning States (ECS) torepresent what visual contents are concerned at each round and what have beenconcerned throughout the Visual Dialogue. ECS is modeled from multimodalinformation and is represented explicitly. Based on ECS, we formulate twointuitive and interpretable rewards to encourage the Visual Dialogue agents toconverse on diverse and informative visual information. Experimental results onthe VisDial v1.0 dataset show our method enables the Visual Dialogue agents togenerate more visual coherent, less repetitive and more visual informativedialogues compared with previous methods, according to multiple automaticmetrics, human study and qualitative analysis.

Modeling Explicit Concerning States for Reinforcement Learning in Visual Dialogue