BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Active Question Answering with Reinforcement Learning
DTSTART:20180122T110000
DTEND:20180122T120000
DTSTAMP:20260502T203157Z
UID:08c1a4f061562405ae3ec4cd3415ef9c5e26eeda9998a9a1febfe038
CATEGORIES:Conferences - Seminars
DESCRIPTION:Massimiliano Ciaramita\, Google Zurich\nAbstract: Humans are c
 apable of solving sophisticated information seeking tasks by interacting w
 ith digital media. For example\, users find answers to complex questions u
 sing search engines\, by formulating multiple queries in iterative\, state
 ful sessions followed by critical assessment and synthesis. Can machine le
 arning be applied to solve similar tasks?\nThis talk presents a framework\
 , called Active Question Answering\, that investigates machine-learned age
 nts that perform information-seeking tasks by using language to interact w
 ith information-providing environments. We start by considering query refo
 rmulation in question answering tasks. Question answering systems frequent
 ly return slightly different answers to different variants of a question. 
 An Active Question Answering agent ideally sits between a user and a black
 -box QA system. This agent learns how to query the QA system optimally on 
 the user's behalf. The agent is trained using Reinforcement Learning to re
 formulate questions and aggregate evidence returned by the QA system in or
 der to infer the best final answer.\nIn an empirical study we show that th
 e agent can learn to outperform the environment\, and other benchmarks\, b
 y a significant margin. We also analyze the  language the agent has learn
 ed by interacting with the QA system. We find that the agent seems to have
  re-discovered basic information retrieval techniques such as tf-idf term 
 reweighting and stemming.  \n \nBio: Massimiliano Ciaramita is a resear
 ch scientist at Google Zurich where he leads a team working on language un
 derstanding. Previously\, he worked at Yahoo! Research and the Italian Nat
 ional Research Council (CNR). He carried out undergraduate studies at the 
 University of Rome "La Sapienza" and obtained Master's and PhD degrees fro
 m Brown University. His main research interests involve language understan
 ding and its applications to web search and conversational technologies. H
 e is also a lecturer at ETH Zurich from 2012\, teaching "Natural Language 
 Understanding".\n 
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR