[Elsnet-list] 2 PhD positions in human-agent interaction in Telecom-ParisTech

Chloé Clavel chloe.clavel at telecom-paristech.fr
Wed Dec 3 16:49:03 CET 2014

We offer two full-time PhD positions. The expected starting date is around early 2015. The duration of the contract is 3 years. The salary is between 21 and 25 k€ per year.
Context of the two PhDs:
The PhDs take part of the european project ARIA - VALUSPA (Affective Retrieval Interface Assistants – using Virtual Agents with Linguistic Understanding, Social skills and Personalised Aspects). The project tackles the development of ECA whose function is to serve as interface for retrieval systems and to provide a more natural answer to user’s requests. For example, the ECA can take the appearance of a novel character to which the user can speak about the content of a novel and about its characters.
The PhDs will take place in the LTCI-CNRS laboratory of Telecom-ParisTech in the GRETA team (http://www.tsi.telecom-paristech.fr/mm/en/themes-2/greta-team/). The first PhD will also rely on a collaboration with the Lattice-CNRS laboratory (http://www.lattice.cnrs.fr/)
Application: We are looking for candidates:
* with a MSc degree in Computer Science or equivalent (a degree with a technical background, e.g., machine learning, signal processing, computer graphics, computational linguistics).
 * with interests in the research fields of social signal processing, machine learning and human-agent interaction.
 * with programming skills: Java 

submit by email at each contact of the corresponding PhD with:
* Curriculum Vitae.
* Mail expressing your interest in the position and your profile relevance (directly in the email body).
* Copy of grades of your MSc degree.
* Contact of a referee and/or recommendation letter.
Incomplete applications will not be processed.

PhD - Verbal alignment strategies in human-agent interactions
Embodied Conversational Agents (ECA) are virtual characters allowing the machine to dialog with humans in a natural way: using not only verbal interaction but also non-verbal interaction (facial expressions, gestures). ECA can take the role of an assistant on sales sites or of tutors in the case of Serious Games. One of the key challenges of human-agent interaction is to maintain user’s engagement in interaction (Sidner et al., 2002). Several strategies can be used to foster this engagement. One of these strategies is ECA alignment on user (Pickering & Garrod, 2000). Alignment can occur at various levels: low-level alignment which consists in the imitation of body postures (Hess et al., 1999) or in the use of a vocabulary close to user’s one; high-level alignment that occur at mental, emotional or cognitive levels. Alignment on user’s opinion or attitude -- also known as affiliation (Stivers, 2008) -- is one example of high-level alignment. 

The PhD follows a previous study carried out at Telecom-ParisTech concerning the analysis of spontaneous human-human conversations in order to define the relevant strategies to foster user’s engagement in human-ECA interaction (Campano, 2014). It will also rely on a collaboration with the Lattice-CNRS and its expertise on the human-machine dialog and on its constraints and linguistic aspects (Landragin, 2013; Vorobyova et al., 2012). It will focus on the development of ECA alignment strategies that will be centered on the verbal modality in a multimodal context (prosody, gesture). The various levels of alignment will be considered. In particular, the PhD will deal with the expression of attitudes (Scherer, 2005; Martin & White, 2003) and with the vocabulary linked to their expression. Alignment on referring expressions (Landragin, 2006) -- here, the way the user and the ECA refers to the targets of the attitude -- will also be studied. The targeted alignment strategies will rely on reasoning methods and on statistical methods (Mairesse, 2010) and will be implemented on the GRETA platform.

Chloé Clavel, associate professor, GRETA team, Télécom ParisTech.
Tel:+33 (0)1 45 81 72 54
E-Mail: chloe.clavel [at] telecom-paristech.fr

Frédéric Landragin, chercheur CNRS, laboratoire Lattice-CNRS.
Tel: +33 (0)1 58 07 66 21
E-Mail: frederic.landragin [at] ens.fr
PhD - Context-sensitive generation of non-verbal behaviours

The Affective Retrieval of Information Assistants, ARIA, are Embodied Conversational Agents with Linguistic Understanding, Social skills, and Personalised Aspects. They should be able to communicate using the same verbal and non-verbal modalities used in human-human interaction, have interpersonal skills akin to those of humans, and adapt to the user in terms of learning their preferences, personality, and manner of interaction. The agents should also be capable of dealing with unexpected situations, such as the user suddenly changing topic or task, the social group changing when a second user arrives, the user interrupting the ECA, or even the user suddenly changing its attitude.
The aim of the PhD is to develop a computational model of the behaviours of the ECA that would convey information at different levels:
·        Interpersonal stance,
·        overall communicative behaviours,
·        emergence of non-verbal alignment with user's behaviour,
·        multimodal response to unexpected situations
The work will make use of an existing ECA platform, Greta (Ochs et al, 2013). In particular we will extend our previous model of multimodal behaviours (Chollet et al., 2014) where we apply sequence mining on data from a corpus to extract frequent sequences for different types of attitude and communicative expressions and to use them as data to generate non-verbal behaviours for the ECAs.
To align the non-verbal behaviour of the ECA with the user, we will use a reinforcement algorithm to update the efficiency of a non-verbal behaviour used to communicate a given intention and/or emotional state. The reinforcement signal will be the achievement of the communicative intention and/or emotional state. The non-verbal behaviour of the ECA will be selected based on its efficiency computed dynamically during the interaction.
 We will simulate various types of behaviour responses to unexpected situations: interruption of behaviour (arising from the stop of current intention) that could be followed by a repair action or an hold (of the gesture), the coarticulation of a behaviour into another one (as the current intention is followed by the instantiation of a new intention), the merge of a behaviour with another one (to adapt the current intention to a new one). Interruption and holding of a behaviour (be a gesture, a facial expression, a gaze behaviour or a torso movement) will be modelled at the level of the behaviour realizer of the virtual agent platform. To coarticulate within another behaviour requires some re-planning to compute which behaviour should appear next; that is which intention that is linked to a next behaviour is triggered.
Catherine Pelachaud, Director of Research CNRS - LTCI, GRETA team, Télécom ParisTech.
Tel:+33 (0)1 45 81 75 93
E-Mail: catherine.pelachaud [at] telecom-paristech.fr

Campano, S., Durand, J. & Clavel, C. (2014) “Comparative analysis of verbal alignment in human-human and human-agent interactions”, In Proceedings of LREC 2014.
Campano, S., Glas, N., Langlet, C., Clavel, C. & Pelachaud, C. (2014) “Alignement par Production d’Hétéro-Répétitions chez un ACA”. In Proceedings of Workshop Affect, Compagnon Artificiel, Interaction.

Mathieu Chollet, Magalie Ochs and Catherine Pelachaud, From Non-verbal Signals Sequence Mining to Bayesian Networks for Interpersonal Attitudes Expression, Intelligent Virtual Agents, IVA'14, pp 120-133, Aug 2014
Chiu, C.C., Marsella, S. (2011): How to train your avatar: a data driven approach to gesture generation. In: Intelligent Virtual Agents, Springer, 127-140, 2011
Y. Ding, M. Radenen, T. Artières, C. Pelachaud, Speech-driven eyebrow motion synthesis with contextual markovian models, ICASSP, USA, 3756-3760, 2013.

Clavel, C., & Richard, G. (2011). “Recognition of acoustic emotion”. Emotion-Oriented Systems, John Wisley, 139-167.

Hess, U., Philippot, P., & Blairy, S. (1999) “8. Mimicry”. The Social Context of Nonverbal Behavior, 213.
Jonsdottir G., Thorisson K., and Nivel E. (2008): Learning smooth, human-like turntaking in realtime dialogue, in H. Prendinger, J. Lester, and M. Ishizuka, editors, Intelligent Virtual Agents, volume 5208 of Lecture Notes in Computer Science, pp. 162–175, Springer Berlin / Heidelberg, 2008.

Landragin, F. (2006) “Visual perception, language and gesture: A model for their understanding in multimodal dialogue systems”. Signal Processing 86.12: 3578-3595.

Landragin, F. (2013) Man-Machine Dialogue. Design and Challenges. Wiley & ISTE Publishing.

Langlet, C. & Clavel, C. (2014) “Modélisation des questions de l’agent pour l’analyse des affects, jugements et appréciations de l’utilisateur dans les interactions humain-agent”, In Actes de TALN 2014, Marseille.
Mairesse, F., Gašić, M., Jurčíček, F., Keizer, S., Thomson, B., Yu, K., & Young, S. (2010) “Phrase-based statistical language generation using graphical models and active learning”. In Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics (pp. 1552-1561). Association for Computational Linguistics.

Martin, J. R. & White, P. R. (2003) The language of evaluation. Palgrave Macmillan.
Ochs, M., Ding, Y., Fourati, N., Chollet, M., Ravenet, B., Pecune, F., Glas, N., Prépin, K., Clavel, C. & Pelachaud, C. (2013) “Vers des Agents Conversationnels Animés Socio-Affectifs”. Interaction Humain-Machine (IHM'13), Bordeaux, France.

Pickering, M. J. & Garrod, S. (2004) Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences 27.2: 169-190.
Prepin, K., Ochs, M. and Pelachaud, C. (2013): Beyond backchannels: co-construction of dyadic stancce by reciprocal reinforcement of smiles between virtual agents. In proceedings of the International Conference CogSci (Annual Conference of the Cognitive Science Society), Berlin, July 2013.
Scherer, K. R. (2005) "What are emotions? And how can they be measured?." Social science information 44.4: 695-729.

Sidner, C. L. & Dzikovska, M. (2002). “Human-robot interaction: Engagement between humans and robots for hosting activities”. In Proceedings of the 4th IEEE International Conference on Multimodal Interfaces (p. 123). IEEE Computer Society.
Stivers, T. (2008) “Stance, alignment, and affiliation during storytelling: When nodding is a token of affiliation”. Research on Language and social interaction 41.1: 31-57.

Vorobyova, A., Benotti, L. & Landragin, F. (2012) “Why do we overspecify in dialogue? An experiment on L2 lexical acquisition”. Proceedings of SemDial 2012, Paris.

More information about the Elsnet-list mailing list