2018/12/28 by Dhanraj Vishwanath, Vishwanath D, Lovekesh Vig +6 · 1 citation
Computer Science · Decision Sciences · Engineering · #AI in Service Interactions #Advanced Bandit Algorithms Research #Artificial Intelligence (cs.AI) #Artificial intelligence #Cognitive Radio Networks and Spectrum Sensing #Computation and Language (cs.CL) #Computer science #Engineering #FOS: Computer and information sciences #Machine Learning (cs.LG) #Operating system #Operations management #Operations research #Reinforcement #Reinforcement learning #Schedule #Scheduling (production processes) #Speech and dialogue systems #Thursday #Topic Modeling #cs.AI #cs.CL #cs.LG
paper · pdf · doi:10.48550/arxiv.1812.11158
published in arXiv (Cornell University) (Cornell University)
arxiv created 2018/12/28 · openalex publication_date 2018/12/28 · arxiv updated 2018/12/31 · openalex created_date 2019/07/30 · openalex updated_date 2026/08/08
In this paper we present Meeting Bot, a reinforcement learning based\nconversational system that interacts with multiple users to schedule meetings.\nThe system is able to interpret user utterences and map them to preferred time\nslots, which are then fed to a reinforcement learning (RL) system with the goal\nof converging on an agreeable time slot. The RL system is able to adapt to user\npreferences and environmental changes in meeting arrival rate while still\nscheduling effectively. Learning is performed via policy gradient with\nexploration, by utilizing an MLP as an approximator of the policy function.\nResults demonstrate that the system outperforms standard scheduling algorithms\nin terms of overall scheduling efficiency. Additionally, the system is able to\nadapt its strategy to situations when users consistently reject or accept\nmeetings in certain slots (such as Friday afternoon versus Thursday morning),\nor when the meeting is called by members who are at a more senior designation.\n