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

We study the problem of balancing effectiveness and efficiency in automatedfeature selection. After exploring many feature selection methods, we observe acomputational dilemma: 1) traditional feature selection is mostly efficient,but difficult to identify the best subset; 2) the emerging reinforced featureselection automatically navigates to the best subset, but is usuallyinefficient. Can we bridge the gap between effectiveness and efficiency underautomation? Motivated by this dilemma, we aim to develop a novel feature spacenavigation method. In our preliminary work, we leveraged interactivereinforcement learning to accelerate feature selection by externaltrainer-agent interaction. In this journal version, we propose a novelinteractive and closed-loop architecture to simultaneously model interactivereinforcement learning (IRL) and decision tree feedback (DTF). Specifically,IRL is to create an interactive feature selection loop and DTF is to feedstructured feature knowledge back to the loop. First, the tree-structuredfeature hierarchy from decision tree is leveraged to improve staterepresentation. In particular, we represent the selected feature subset as anundirected graph of feature-feature correlations and a directed tree ofdecision features. We propose a new embedding method capable of empoweringgraph convolutional network to jointly learn state representation from both thegraph and the tree. Second, the tree-structured feature hierarchy is exploitedto develop a new reward scheme. In particular, we personalize reward assignmentof agents based on decision tree feature importance. In addition, observingagents' actions can be feedback, we devise another reward scheme, to weigh andassign reward based on the feature selected frequency ratio in historicalaction records. Finally, we present extensive experiments on real-worlddatasets to show the improved performance.

Interactive Reinforcement Learning for Feature Selection with Decision Tree in the Loop