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

When humans solve complex problems, they typically create a sequence of ideas(involving an intuitive decision, reflection, error correction, etc.) in orderto reach a conclusive decision. Contrary to this, today's models are mostlytrained to map an input to one single and fixed output. In this paper, weinvestigate how we can give models the opportunity of a second, third and$k$-th thought. Taking inspiration from Hegel's dialectics, we propose theconcept of a thought flow which creates a sequence of predictions. We present aself-correction mechanism that is trained to estimate the model's correctnessand performs iterative prediction updates based on the correctness prediction'sgradient. We introduce our method at the example of question answering andconduct extensive experiments that demonstrate (i) our method's ability tocorrect its own predictions and (ii) its potential to notably improve modelperformances. In addition, we conduct a qualitative analysis of thought flowcorrection patterns and explore how thought flow predictions affect human userswithin a crowdsourcing study. We find that (iii) thought flows enable improveduser performance and are perceived as more natural, correct, and intelligent assingle and/or top-3 predictions.

Thought Flow Nets: From Single Predictions to Trains of Model Thought