BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Structured Deep Learning for Dialogue Management - Kai Yu\, Shangh
 ai Jiao Tong University
DTSTART:20180219T120000Z
DTEND:20180219T130000Z
UID:TALK101266@talks.cam.ac.uk
CONTACT:Anton Ragni
DESCRIPTION:In this talk\, I will first introduce the works on structured 
 deep learning for speech and language processing at the SpeechLab of Shang
 hai Jiao Tong University. Then\, I will focus on a recent work on structur
 ed deep reinforcement learning. When dialogue domain changes dynamically\,
  e.g. a new previously unseen concept (or slot) which can be then used as 
 a database search constraint is added\, or the policy for one domain is tr
 ansferred to another domain\, the dialogue state space and action sets bot
 h will change. Therefore\, the model structures for different domains have
  to be different. This makes dialogue policy adaptation/transfer challengi
 ng. Here a multi-agent dialogue policy (MADP) is proposed to tackle these 
 problems. MADP consists of some slot-dependent agents (S-Agents) and a slo
 t-independent agent (G-Agent). S-Agents have shared parameters in addition
  to private parameters for each one. During policy transfer\, the shared p
 arameters in S-Agents and the parameters in G-Agent can be directly transf
 erred to the agents in extended/new domain. Simulation experiments showed 
 that MADP can significantly speed up the policy learning and facilitate po
 licy adaptation.
LOCATION:Department of Engineering - LR3B Lecture Room
END:VEVENT
END:VCALENDAR