Open Peer Review. Open Publishing. Open Access. Open Discussion. Open Directory. Open Recommendations. Open API. Open Source.
Knapsack Constrained Contextual Submodular List Prediction with Application to Multi-document Summarization
Jiaji Zhou, Stephane Ross, Yisong Yue, Debadeepta Dey, J. Andrew Bagnell
May 01, 2013 (modified: May 01, 2013)ICML 2013 Inferning submissionreaders: everyone
Abstract:We study the problem of predicting a set or list of options under knapsack constraint. The quality of such lists are evaluated by a submodular reward function that measures both quality and diversity. Similar to DAgger approach , by a reduction to online learning, we show how to adapt two sequence prediction models to imitate greedy maximization under knapsack constraint problems: CONSEQOPT and SCP. Experiments on extractive multi-document summarization show that our approach outperforms existing state-of-the-art methods.
Enter your feedback below and we'll get back to you as soon as possible.