Search

Addressee detection is a fundamental task for seamless dialogue management and turn taking in human-agent interaction. Though addressee detection is implicit in dyadic interaction, it becomes a challenging task when more than two participants are involved. This article proposes multiple addressee detection models based on smart feature selection and focus encoding schemes. The models are trained using different machine learning and deep learning algorithms. This research work improves existing baseline accuracies for addressee prediction on two datasets. In addition, the article explores the impact of different focus encoding schemes in several addressee detection cases. Finally, an implementation strategy for addressee detection model in real-time is discussed.

International audience; Addressee detection is one of the most fundamental tasks for seamless dialogue management and turn taking in human-agent interaction. Whereas addressee detection is implicit in dyadic interaction, it becomes a challenging task in multiparty interactions when more than two participants are involved. Existing research works employ either rule-based or statistical approaches for addressee detection. However, most of these works either have been tested on a single data set or only support a fixed number of participants. In this article, we propose a model based on generic features to predict the addressee in data sets with varying number of participants. The results tested on two different corpora show that the proposed model outperforms existing baselines.

Background Telomerase reverse transcriptase promoter (pTERT) mutation is a dominant altered telomere maintenance mechanism in primary glioblastomas (GBMs). Objective The aim of this study was to correlate pTERT mutations with clinico-histological features and other molecular markers (p53 protein-expression, ATRX protein-expression, IDH mutations, EGFR gene amplification and MGMT methylation) in adult GBMs. Materials and methods Evaluated for histological patterns, p53 and ATRX protein expression by immunohistochemistry (IHC), IDH mutations by IHC followed by sequencing in IHC negative cases, EGFR gene amplification by fluorescence in situ hybridization, MGMT promoter methylation by methylation-specific PCR and pTERT mutation by sequencing. Results A total of 155 adult supratentorial GBMs [age-range 20-80 years] formed study cohort. 15.6% were IDH1R132 mutated, none were IDH2R172 mutated and 27% were EGFR amplified. 43% were MGMT methylated and were more common with IDH-mutation (mIDH) than EGFR amplification. 90% of mIDH (but no EGFR amplified) cases showed ATRX-loss. 43.5% were pTERT mutated (C228T was the commonest type) and were mutually exclusive with ATRX-loss. 14% of mIDH and 42% of EGFR amplified cases showed pTERT mutation, the latter was more commonly pMGMT unmethylated (63.6%). Conclusions 43.5% of the GBMs showed pTERT mutation (C228T was commonest; 72%). pTERT mutations were mutually exclusive with ATRX protein loss, more commonly associated with IDH wild type and EGFR amplified GBMs.

International audience; Addressee detection is an important challenge to tackle in order to improve dialogical interactions between humans and agents. This detection, essential for turn-taking models, is a hard task in multiparty conditions. Rule based as well as statistical approaches have been explored. Statistical approaches, particularly deep learning approaches, require a huge amount of data to train. However, smart feature selection can help improve addressee detection on small datasets, particularly if multimodal information is available. In this article, we propose a statistical approach based on smart feature selection that exploits contextual and multimodal information for addressee detection. The results show that our model outperforms an existing baseline.

Automatic dialogue act annotation of speech utterances is an important task in human-agent interaction in order to correctly interpret user utterances. Speech utterances can be transcribed manually or via Automatic Speech Recognizer (ASR). In this article, several Machine Learning models are trained on manual and ASR transcriptions of user utterances, using bag of words and n-grams feature generation approaches, and evaluated on ASR transcribed test set. Results show that models trained using ASR transcriptions perform better than algorithms trained on manual transcription. The impact of irregular distribution of dialogue acts on the accuracy of statistical models is also investigated, and a partial solution to this issue is shown using multimodal information as input.

INTRODUCTION: Pilocytic astrocytomas (PCA) are characterised by BRAF fusions and V600E mutation. MATERIAL and METHODS: FFPE tissues of PCA diagnosed during 2011-17 were evaluated for BRAFV600E mutation by Sanger sequencing and KIAA1549:BRAF fusion transcripts (KIAA1549:BRAF 16-9, KIAA1549:BRAF 15-9 and KIAA1549:BRAF 16-11) by reverse transcriptase polymerase chain reaction. RESULTS: 272 cases of PCA of age range 1-46 years (≤14 years:176; 15-25 yrs:76, 26-39yrs:19 and >39yrs:1) formed the sample. BRAFV600E and fusions were mutually exclusive. 25 of 234 (10.7%) were BRAFV600E mutated [cerebellum: 9/90(10%), suprasellar: 5/30(17%), CH: 4/27(15%), brain stem: 2/14(14%) and thalamic: 2/13(15%) 4th ventricle: 1/10 (10%) and 3rd ventricle: 2/7 (28.6%). Whereas 75 of 224 (33.5%) were BRAF fusion [KIAA1549-BRAF 16-9 (n: 52), 15-9 (n:15) & 16-11 (n:8)] positive. 40% of cerebellar (36/91), 37% of suprasellar (10/27), 43% of brain stem (6/14), 36% of spinal (5/14), 21% of CH (5/24), 21% of ON (4/19), 27% of thalamic (3/11), 55.6% of 4th ventricle (5/9) and 50% of pineal (1/2) were fusion positive. None of the 19 ON cases showed BRAFV600E. It was also not detected in 25 yrs age; but was seen in 15.4% (10/65) of cases in 15-25yrs. BRAF fusions were more common in ≤14 yrs (57/145; 39%). Interestingly, none of cases >30 years showed BRAF alterations. Two cases were associated with NF-1 both were negative for BRAF alterations. CONCLUSIONS: BRAF fusions are common than BRAFV600E in PCA across all locations and interestingly both are extremely rare >30years.

In the last decades, the industry has profoundly integrated the use of digital resources in their production process. However, these assets are rarely re-used for the training of the users, operators and technicians that have to interact with these objects. Furthermore, although training and learning environments are classical applications of virtual reality, the design of these environments is generally ad hoc, i.e. dedicated to specific operations on specific objects, hence requiring the intervention of programmers whenever a modification of the pedagogical scenario is required. In this article, we propose a methodology to design adaptable virtual environments, by separating the role of the different protagonists that play a part in the creation of learning environments. In particular, its goal is to allow the teachers to implement different scenarios according to the level of the trainees and to the pedagogical objectives without the intervention of computer scientists. An example of adaptable wind turbine environment is shown, with three different learning situations: simulator, safety training and preventive maintenance training.

International audience

Virtual agents are a real asset in collaborative virtual environment for training CVET as they can replace missing team members. Collaboration between such agents and users, however, is generally limited. We present here a whole integrated model of CVET focusing on the abstraction of the real or virtual nature of the actor to define a homogenous collaboration model. First, we define a new collaborative model of interaction. This model notably allows to abstract the real or virtual nature of a teammate. Moreover, we propose a new role exchange approach so that actors can swap their roles during training. The model also permits the use of physically based objects and characters animation to increase the realism of the world. Second, we design a new communicative agent model, which aims at improving collaboration with other actors using dialog to coordinate their actions and to share their knowledge. Finally, we evaluated the proposed model to estimate the resulting benefits for the users and we show that this is integrated in existing CVET applications. Copyright © 2014 John Wiley & Sons, Ltd.

Virtual environments for human learning enable one or more users to interact with virtual agents in order to perform their tasks. This collaboration necessitates that the members of the team share a set of beliefs and reason about resources, plans and actions to be implemented. This article introduces a new multiparty coordination model allowing several virtual and human agents to dialogue and reason about the tasks that the user must learn. The proposed model relies on a shared plan based approach to represent the beliefs of the team members. The management of the multiparty aspect makes it possible to differentiate the behaviors to be produced according to the type of receiver of a communication: recipient or listener. Finally, in the context of learning a procedural activity, a study examines the effect of our multiparty model on a learner. Results show that the use of proactive pedagogical agents with multiparty competencies boosts the construction of common beliefs.

The EAST (Scientific and technical learning environments) project aims at stimulating the interest of young people for science through virtual reality environments, based on industrial assets. Although training and learning environments are classical applications of virtual reality, the design of these environments is generally ad hoc, hence requiring the intervention of programmers whenever a modification of the pedagogical scenario is required. In this paper, we propose a methodology to design virtual environments which can be adapted by teachers to implement different scenarios according to the level of the trainees and to the pedagogical objectives. Current demonstrators include a windmill with three different learning situations: simulator, safety training and preventive maintenance training.

International audience; The coordination is an essential ingredient for the human-agent teamwork. It requires team members to share knowledge to establish common grounding and mutual awareness among them. In this paper, we propose a behavioral architecture C 2 BDI that allows to enhance the knowledge sharing using natural language communication between team members. We define collaborative conversation protocols that provide proactive behavior to agents for the coordination between team members. We have applied this architecture to a real scenario in a col-laborative virtual environment for training. Our solution enables users to coordinate with other team members.

In Collaborative Virtual Environments (VEs) for Training, users have to learn how to perform a collaborative task and also how to coordinate with teammates’ activities. Efficient coordination requires teammates to exchange information about their beliefs, goals and plans. The collaborative-conversational BDI agent (C2BDI) endows virtual agents with first, deliberative capabilities about the interdependency of their activities, and second, with task-oriented conversational capabilities that support multiparty spoken dialogues helping them to coordinate their activities with teammates [2]. This proposed solution has been used in two virtual reality applications: a real training scenario [1] and an application dedicated to scientific experiments [2]. The main motivations of this last was to control the characteristics of the collective activity and to be more extensible.

National audience; Cet article propose une approche innovante de la gestion et de l'échange de connaissances entre utilisateurs et agents autonomes dans les Environnements Virtuels de Formation Collaboratifs. Afin de faciliter à la fois le dialogue et la réalisation de tâches collaboratives au sein d'une équipe pouvant être composée d'utilisateurs et d'agents autonomes, nous introduisons une entité commune à ces derniers : le Shell. Cette entité regroupe la gestion des connaissances et le contrôle du mannequin qui lui est attaché. Les agents autonomes, basés sur une architecture nommée C-BDI, exploitent les connaissances du Shell auquel ils sont liés afin d'alimenter leur processus de prise de décision et d'enrichir leur dialogue, basé sur une approche Information-State. Les utilisateurs peuvent également accéder aux connaissances de leur Shell respectif à travers une in- terface utilisateur.

Many Virtual Reality (VR) applications, such as Virtual Learning Environments or Interactive Virtual Tours, are based on a rich semantic description of the environment and tasks that users have to perform. These applications are built upon Virtual Environments (VEs) in which artificial agents act autonomously while interacting in realtime with users. Semantic modelling of a VR environment makes it possible the knowledge-driven access from the description of VEs that simplifies the development of VR applications. It eases the development of these types of applications. Semantic modelling should provide a consistent representation of the following aspects: 1) The simulated world, its structure and the behavior of its entities, 2) Interactions and tasks, that users and agents can perform in the environment, 3) Knowledge items, that autonomous agents can use for decision-making or for communication with users. This paper presents MASCARET, a model-based approach, for the design of semantic VR environments. This approach is based on the Unified Modeling Language (UML). In this approach, UML is used to provide a knowledge-driven access to the semantic contents of the VE and not for code generation, as in classical software development process. Interests of a UML-based approach are that its metamodel covers different views of the semantic modelling: ontology, structure, behaviors, interactions, activities. It is also an extensible language that can be specialized to provide formal operational semantics. We also present how MASCARET can be used to develop content-rich interactive applications that can be deployed over various VR platforms. Finally, we discuss the benefits of such a metamodel-based approach and show how the multi-layer semantic model can be used in different VR applications, in which adaptive behaviors of artificial agents acting within complex environments have to be simulated.

International audience; Recent cultural heritage applications have been based on rich-content virtual environment (VE), in which virtual humans can communicate with visitors and other agents using natural language (NL). The conceptualisation of these dialogues are dependent on the contents of the application. Hence, we propose to use the semantic modelling of the VE and the agents' activities for the conceptualisation of the dialogue. Meta-level semantic information are used as arguments in NLU/NLG rules. The advantage of this approach is that the dialogue rules are independent from the contents of the application and have clear semantics. We applied these principles to develop Brest'Coz, an interactive virtual tour for the learning of shipbuilding techniques used in France in early 18 th century.