Img

DSTC6

Dialog System Technology Challenges
Long Beach, USA, December 10, 2017

Workshop Program

December 10, 2017
@ Hyatt Regency Long Beach, Seaview Ballroom

08:30-09:00Registration
   
Opening
09:00-09:10Welcome Remarks from DSTC Steering CommitteeJason Williams (Microsoft Research)
09:10-09:15DSTC6 OverviewSeokhwan Kim (Adobe Research)
   
Track 1: End-to-End Goal Oriented Dialog Learning
09:15-09:35Track OverviewJulien Perez (Naver Labs Europe)
09:35-09:55Extended Hybrid Code Networks for DSTC6 FAIR Dialog DatasetJiyeon Ham, Soohyun Lim, Kee-Eung Kim
09:55-10:15Modeling Conversations to Learn Responding Policies of
E2E Task-oriented Dialog System
Ziwei Bai, Bo Yu, Guangyu Chen,
Baoxun Wang, Zhuoran Wang
   
10:15-10:30Coffee break
   
Invited Talk 1
10:30-11:10 Dialogue as multitask learning and question answering Dr. Richard Socher (Salesforce)
11:10-11:20Q&A 
   
Track 2: End-to-End Conversation Modeling
11:20-11:40Track OverviewChiori Hori (MERL)
11:40-12:00The MSR-NLP System at Dialog System Technology Challenges 6Michel Galley, Chris Brockett, Bill Dolan, Jianfeng Gao
   
Invited Talk 2
12:00-12:40Visual Dialog: Towards AI agents that can see, talk, and actDr. Dhruv Batra (Georia Tech/FAIR)
12:40-12:50Q&A 
   
12:50-14:10Lunch Break & Sponsor Talks
   
Track 3: Dialog Breakdown Detection
14:10-14:30Track OverviewRyuichiro Higashinaka (NTT)
14:30-14:50Dialogue Breakdown Detection based on
Estimating Appropriateness of Topic Transition
Hiroaki Sugiyama
14:50-15:10RSL17BD at DBDC3: Computing Utterance Similarities
based on Term Frequency and Word Embedding Vectors
Sosuke Kato , Tetsuya Sakai
15:10-15:30How Generic Can Dialogue Breakdown Detection Be?
The KTH entry to DBDC3
Jose Lopes
   
15:30-15:45Coffee Break & Flash Talk by Poster Presenters
   
Poster Session
15:45-17:15Poster Presentations:
 End-to-End Recurrent Entity Network for Entity-Value Independent Goal-Oriented Dialog Learning
Chien-Sheng Wu, Andrea Madotto, Genta Winata, Pascale Fung
 Learning Dynamic Memory Network with Two Views
Chang-Uk Shin, Jeong-Won Cha
 End-to-end memory networks with word-abstraction and contextual-numbering for goal oriented task
Asuka Sakai, Hongjie Shi, Takashi Ushio, Mitsuru Endo
 End-to-End Goal Oriented Dialog Learning Based On Memory Network
Byoungjae Kim, KyungTae Chung, Jeongpil Lee, Jungyun Seo, Myoung-Wan Koo
 Quantized-Dialog Language Model for Goal-Oriented Conversational Systems
Chulaka Gunasekara, David Namahoo, Lazaros Polymenakos, Jatin Ganhotra, Kshitij Fadnis
 A Knowledge Enhanced Generative Conversational Service Agent
Yinong Long, Jianan Wang, Zhen Xu, Zongsheng Wang, Baoxun Wang, Zhuoran Wang
 Comparative Analysis of Word Embedding Methods for DSTC6 End-to-End Conversation Modeling Track
Bairong Zhuang, Wenbo Wang, Zhiyu Li, Chonghui Zheng, Takahiro Shinozaki
 Sequence Adversarial Training and Minimum Bayes Risk Decoding for End-to-end Neural Conversation Models
Wen Wang, Yusuke Koji, Bret Harsham, Takaaki Hori, John Hershey
 End-to-end Character-Level Dialogue Breakdown Detection with External Memory Models
Atsushi Saito, Taichi Iki
 Dialogue Breakdown Detection using Hierarchical Bi-Directional LSTMs
Zeying Xie, Guang Ling
 Attention-based Dialog Embedding for Dialog Breakdown Detection
Chanyoung Park, Kyungduk Kim, Songkuk Kim
 Dialogue Breakdown Detection Considering Annotation Biases
Junya Takayamy, Eriko Nomoto, Yuki Arase
   
Closing
17:15-17:55DSTC6 Review/DSTC7 Track ProposalsKoichiro Yoshino (NAIST)
17:55-18:00Wrap-up TalkChiori Hori (MERL)

December 11, 2017

Optional Excursion
13:30-16:30Lab tour at University of Southern California (optional)

For lab tour participants: Please gather at,
12015 Waterfront Drive Playa Vista, CA 90094-2536
by 13:15. We don't provide any transportation, please share ride service.

Supporting Organizations

The DSTC6 workshop is sponsored by Adobe Research, Mitsubishi Electric Research Laboratories, and Nara Institute of Science and Technology, and endorsed by International Speech Communication Association (ISCA). The workshop proceedings will be available in the ISCA archive.