Frédéric Béchet is a researcher in the field of Speech and Natural Language Processing at the Laboratoire d'Informatique Fondamentale (LIF) of Marseille.
After studying Computer Science at the Aix-Marseille II University, he obtained his PhD in Computer Science in 1994 from the University of Avignon, France. Since then he worked at the Ludwig Maximilian University in Munich, Germany, as a Professor Assistant at the University of Avignon, France, as an invited professor at AT&T Research Shannon Lab in Florham Park, New Jersey, USA.
He is currently a full Professor of Computer Science at the Aix Marseille Université in France since 2009.
In terms of managerial activities, Frédéric Béchet is currently at the head of the research departement LIS - UMR7020 https://www.lis-lab.fr/ .
My research activities are focused on automatic language processing models for Spoken Language Understanding. By dealing with spoken language, I’m working at the crossing of Natural Language Processing on one hand and Automatic Speech Recognition on the other hand.
The main research topics that I studied during my academic career are :
These topics have been studied through three applicative frameworks: automatic spoken dialog systems; speech analytics; linguistic processing of non-canonical texts. One focus of my work is to combine academic research to applicative frameworks defined through collaborative projects with industrial partners. These projects are a unique opportunity to work on large "realistic" datasets in order to develop and evaluate new language processing models. These models are mostly corpus-based models with various machine learning methods integrating linguistically motivated resources.
After studying Computer Science at the University of Marseille I obtained my PhD in 1994 at the University of Avignon with Professor Henri Meloni on Language Processing for Automatic Speech Recognition.
After this PhD I did a post-doctoral research contract of 16 months at the Computational Linguistics Laboratory (CIS) of the Ludwig Maximilian University of Munich, Germany. This experience allowed me to precise the line of research I still follow today, at the intersection of Natural Language Processing one hand, and Speech Processing on the other hand.
I obtained a Professor Assistant position at the University of Avignon in September 1995. I joined Professor de Mori team, working on speech processing for Spoken Dialog Systems.
I went on a sabbatical leave for one year in 2001 which I spent as an Invited Professor at the AT&T Shannon Labs in New Jersey, USA. I worked in Allan Gorin team on the Spoken Dialog System “How May I Help You?”.
I obtained my « habilitation à diriger des recherches » in 2007 and a Full Professor position at the Aix Marseille University in 2009. I joined the NLP team of the Laboratoire d’Informatique Fondamentale (LIF).
My first area of research, during my PhD, was the application of analytical models, knowledge based, for Automatic Speech Recognition. Former student of Alain Colmerauer, this line of work was very influenced by logic programming, inference and Artificial Intelligence. After my PhD I started working with machine learning corpus-based approaches for speech processing.
Since my arrival in the Natural Language Processing team (TALEP) of the Computer-Science lab (LIF) of Marseille, I work on the combination of efficient Machine Learning approaches with rich linguistic resources. During these 5 past years I have been interested in multimodal approaches language processing. By considering language in several modalities (image, audio, text), I try to combine through joint processing different analyses that can be done in a multimodal document such as a video.
Frédéric Béchet is a researcher in the field of Speech and Natural Language Processing. His research activities are mainly focused on Spoken Language Understanding for both Spoken Dialogue Systems and Speech Mining applications. After studying Computer Science at the University of Marseille, he obtained his PhD in Computer Science in 1994 from the University of Avignon, France. Since then he worked at the Ludwig Maximilian University in Munich, as a Professor Assistant at the University of Avignon, as an invited professor at ATT Research Shannon Lab in Florham Park, New Jersey.
Frédéric Béchet is currently a full Professor of Computer Science at the Aix Marseille University, and a member of the Natural Language Processing research group of the Laboratoire d’Informatique de Marseille (LIF-CNRS).
Frédéric Béchet is the author/co-author of over 100 refereed papers in journals and international conferences and hold two patents.
He is an Associate Editor for IEEE Signal Processing Letter since 2012, has served on the reviewing committees of several international conferences (ICASSP, Interspeech, ASRU, HLT, EMNLP) and has been an invited reviewer for several journals including : Speech Communication, IEEE Signal Processing Letters, IEEE Transactions on Speech and Audio Processing, Traitement Automatique des Langues.
Frédéric Béchet was an elected member of the IEEE Speech and Language Processing Technical Committee and is currently vice-president of the board of the French Natural Language Processing association ATALA.
Frédéric Béchet has been involved in many French and European research programs in the fields of Speech Processing and Spoken Dialog Systems : FP5 SMADA STREP, FP6 LUNA STREP, FP6 PASCAL NoE, ANR EPAC, ANR SEQUOIA, ANR EDYLEX, ANR DECODA, ANR PERCOL, ANR ASFALDA, ANR ORFEO, ANR DATCHA.
He was the Coordinator of several French programs such as PERCOL, DECODA and DATCHA, as well as Principal Investigator for the Aix Marseille University for a US DARPA funded project (BOLT 2012-2015).
In terms of managerial activities, Frédéric Béchet is currently at the head of the Computer Science Research Federation of the Aix Marseille University (Fédération de Recherche en Informatique et Interactions d’Aix-Marseille, FRIIAM - FR3513). 3
My research activities are mainly focused on automatic language processing models for Spoken Language Understanding (SLU). By dealing with spoken language, I’m working at the crossing of Natural Language Processing (NLP) on one hand and Automatic Speech Recognition (ASR) on the other hand.
The main research topics that I studied during my academic career are :
These topics have been studied through three applicative frameworks: automatic spoken dialog systems; speech analytics ; linguistic processing of non-canonical texts. One focus of my work is to combine academic research to applicative frameworks defined through collaborative projects with industrial partners. These projects are a unique opportunity to work on large "realistic" datasets in order to develop and evaluate new language processing models.
These models are mostly corpus-based models with various machine learning methods integrating linguistically motivated ressources.
Along these lines I have participated to many collaborative projects with international academic partners (e.g. LIMSI and Paris7-Alpage in France; University of Trento, University of Sheffield, RWTH Aachen University in Europe; Columbia University, Washington University in USA) and industrial partners leaders in their fields related to Spoken Language Processing (France-Telecom Orange in France ; Loquendo and Teleperformance in Italy ; AT&T Labs and SRI International in USA).
In addition to these collaborative projects, I have participated to many evaluation programs on a large range of NLP tasks in order to validate the models and methods proposed by comparing them to other teams results, such as :
My current research activities are focused on two main topics: robust linguistic analysis and multimodal language processing, both applied to interactive language.
By interactive language I consider not only spoken dialogs, but also all computer-mediated communication such as chat, comments on blogs, tweets, etc.
My work on Robust linguistic analysis is about the adaptation of parsing models (syntactic, semantic, discourse) to process noisy text either produced by an automatic system (Automatic Speech Recognition or Machine Translation systems) or representing non-canonical language (spontaneous speech, social-media data). The use of linguistically motivated models for performing multi-level parsing, in addition to supervised and unsupervised machine learning methods, aims to provide a generic representation of a message that can be used to perform a large range of language understanding tasks, from
ASR error correction as in the DARPA funded BOLT project (2011-2015) where clarification dialogs are used to correct a message prior to translation, to abstractive summary generation in the European project SENSEI (2014-2016) dedicated to extract knowledge from large archives of spoken and text conversations.
Since my arrival at the Aix Marseille University, in September 2009, I collaborate with the Machine Learning and Multimedia processing team (QARMA) of my lab, leading to an extension of my research areas to multimodal language processing in video documents. This collaboration lead to the participation to the French ANR-DGA scientific challenge REPERE (2010-2014) which was about multimodal person recognition in broadcast video documents. I was the coordinator of a 3-years funded collaborative project with the University of Avignon, the University of Lille and France-Telecom Orange Labs in order to participate to this challenge.
Our consortium was ranked first on the main task at the last 2014 evaluation. My main contribution in this direction of research is to consider the problem as a multimodal understanding problem where the different modalities are integrated as early as possible in the understanding process instead of the standard late fusion approach. In particular scene analysis was integrated into the speaker role recognition process prior to speaker clustering in the person identification process for the REPERE challenge.
This work is currently extended through two projects, one funded by the French DGA on multimodal language understanding, the other one funded by the French Intiative of Excellence A*MIDEX on video recommendation.