Institut des Technologies Multilingues & Multimédias de l’Information
Institute for Multilingual & Multimedia Information
Institut für Multilinguale & Multimediale Informationsverarbeitung

3. Program

jeudi 3 avril 2014

Toutes les versions de cet article : [français] [français]

GIF - 2.9 ko
Errare Humanum est

Wednesday, 20th of November 2013

14h00-18h00 Camomile International Advisory committee (AdCom) meeting (Chair Joseph Mariani)

20:30 Dinner at the castle

Day 1, Thursday, 21st of November 2013

8:30-8:45 Welcome/Presentation of the workshop

8:45 Session 1 Modelisation of Errors by machine
(Chair : Yannick Estève)

8:45-9:40 Keynote Julia Hirschberg, Svetlana Stoyanchev, Philipp Salletmayer, Jingbo Yang, Alex Liu, Eli Pincus

Identifying and Responding to Errors in Spoken Dialogue Systems

9:40-10:05 Maja Popović

Automatic error analysis of machine translations

10:05-10:30 Martin Sundermeyer, Volker Steinbiss, Hermann Ney

Direct Observation of Pruning Errors - A Search Analysis Tool

 10:30-10:55 Atsunori Ogawa, Takaaki Hori, Atsushi Nakamura and Takanobu Oba

Recognition rate estimation based on error type classification and its applications

 10:55-11:15 Discussion

 11:15-12:30  Visit of Jean-Jacques Rousseau’s Park

12:30-13:30 Lunch

 13:45 Session 2 Complex errors, errors from multiple sources
(Chair : Chris Cieri)

13:45-14:40 Keynote Josef Kittler

To fuse or not to fuse

14:40-15:05 Juliette Kahn, Ilya Oparin, Olivier Galibert

Evaluation is not just a score

15:05-15:30 Hervé Bredin, Pierrick Bruneau, Anh-Phuong Ta, Claude Barras

Visualizing multimodal person recognition errors : a REPERE use case of the Camomile project

15:30-15:55 Mirjam Ernestus

Errors induced by pronunciation variants typical of casual speech

15:55-16:15 Discussion

16:15-16:30 Break

16:30 Session 3 Impact of errors
(Chair : Laurent Besacier)

16:30-17:25 Keynote Lucia Specia

Is this translation fit for purpose ? Predicting quality versus predicting errors

17:25-17:50 Sophie Rosset, Daniel Luzzati, Cyril Grouin, Ioana Vasilescu, Martine Adda-Decker, Eric Bilinski, Nathalie Camelin, Juliette Kahn, Carole Lailler, Lori Lamel

Human annotation of ASR error regions : Is ’gravity’ a sharable concept for human annotators ?

17:50-18:15 Catherine Kobus et Jean Senellart

Machine translation error categorization by post-editions automatic analysis

18:15-18:40 Benoit Favre, Kyla Cheung, Siavash Kazemian, Adam Lee, Yang Liu, Cosmin Munteanu, Ani Nenkova, Dennis Ochei, Gerald Penn, Stephen Tratz, Clare Voss, Frauke Zeller

Automatic Human Utility Evaluation of ASR Systems : Does WER Really Predict Performance ?

18:40-19:00 Discussion

20:00 Concert (1st part)

Benoît SAROCCHI and musicians traditional Corsican songs

20:30 Gala dinner

22:00 Concert (2nd part)

22:30 Dessert

23:00 Open Bar

Day 2, Friday, 22nd of November 2013

9:00 Session 4 Errors by machines and humans
(Chair : Ioana Vasilescu)

9:00-9:55 Keynote Sarah Hawkins

Illusions, contexts and domains of analysis in assessing errors in speech processing

9:55-10:20 Michel Simard, Pierre Isabelle, George Foster

Error Analysis of Error Corrections in Machine Translation

10:20-10:45 Martin Cooke, Maria Luisa Garcia Lecumberri, Mate Toth

Noise-induced slips of the ear

10:45-11:10 Break & check out

11:10-11:35 Odette Scharenborg

Comparing human and automatic speech recognition

11:35-12:00 Bernd T. Meyer

What’s the difference ? Quantifying errors in human and automatic speech recognition

12:00-12:20 Discussion

12:20-13:35 Lunch

13:35 Session 5 Error Analysis
(Chair : Liz Shriberg)

13:35-14:30 Keynote Mark Liberman

Machine learning from machine learning errors : Work in progress on parsing and phonetic alignment

14:30-14:55 Sebastian Stüker

The Influence of Topic and Domain Specific Words on WER - And Can We Get the User in to Correct Them ?

14:55-15:20 Natalie Kübler, François Yvon , Guillaume Wisniewski

Human Errors and Automatic Errors in Machine Translations. What are the Differences ?

15:20-15:35 Break

15:35-16:00 Dominic Telaar, Mark Erhardt, Tanja Schultz (presented by Tim Schlippe)

Error blaming based on Decoding Output

16:00-16:25 Martine Adda-Decker, Barbara Schuppler, Lori Lamel, Juan Andrés Morales-Cordovilla, Gilles Adda

What we can learn from ASR errors about low-resourced languages : a case-study of Luxembourgish and Austrian

16:25-16:45 Discussion

16:45-17:00 Break

17:00-17:30 Final discussion
(Chairs : Gilles Adda and Joseph Mariani)

17:00-17:10 Summary by Ilya Oparin
17:10-17:20 Perpective by Daniel Luzzati
17:20-17:30 What’s next ? Gilles Adda

18:00 Bus departure

SPIP | squelette | | Plan du site | Suivre la vie du site RSS 2.0