Servizio Comunitario di Informazione in materia di Ricerca e Sviluppo - CORDIS


ADAMACH Sintesi della relazione

Project ID: 22593
Finanziato nell'ambito di: FP6-MOBILITY
Paese: Italy

Final Activity Report Summary - ADAMACH (Adaptive and Meaning Machines)

The research goal of the ADAMACH project was interdisciplinary and its research team members' background ranges from electrical engineering, computer science, psychology, applied mathematics. The Marie Curie Fellows come from an international experience (United Kingdom, United States, Russia, Czech Republic) and they are supported by the Signals and Interactive Systems lab (formerly AMI2 ) founded by the Marie Curie Excellence Leader, Prof. Riccardi (University of Trento, Italy).

The objective of the ADAMACH project was to create machines that could interact with people and help them perform tasks in their everyday lives. In order to accomplish that, the Marie Curie team has investigated how verbal (e.g. words) and non-verbal (e.g. personality) information can be decoded from the speech signal. Such words are the basic bricks to compute the meaning of utterances spoken from people. A cognitive-inspired approach to the interaction model has been addressed to model the decisions made by the machine. Last but not least, the 'personal' component of human-human interactions has been studied to understand the role of personality traits. The validation of such mathematical models has required an advanced experimental infrastructure to validate human-machine interactions with real users.

Main scientific results:
- mathematical models for recognising and understanding conversational speech;
- machine learning framework for modelling human-machine interactions;
- evaluation of personal aspects of human-human communication;
- technological platform to design and evaluate multilingual conversational agents.

The ADAMACH project has created machine prototypes that can interact naturally with real users. Such machines are able to learn automatically from past conversation with real users, what they have said and how they interacted.

The research carried out in the ADAMACH project has been published in top international conferences and journals. Its innovation stems from the state-of-the-art models and technology as well as solid evaluation protocol on real-users. Its members have received awards (e.g. IEEE SPS best paper award), invited to present their prototypes in research and technology forum and supported the W3C standardisation of multimodal human-machine interaction platforms. Important research and technological partnerships have been signed with leading companies and research labs around world. A spinoff has been started to support the development and commercialisation of conversational agent technology.

The ADAMACH project has had an impact on European policy in two ways. First young and senior researchers have had the opportunity to work on a high-risk research project in a truly interdisciplinary and international environment. They have had the opportunity to access the European network of research excellence in the human language technology sector. Secondly, while the conversational agent technology is still in its infancy the ADAMACH project has created a network with major technology player and potential user-end community in order to seek business opportunities in sector such as telecommunications.

The SIS lab along with his international network of scientific and technology partners has provided the ideal environment to Marie Curie fellows to evaluate their research ideas with high societal impact. The Marie Curie fellows have had many opportunities to connect to potential employers both in the academia and private research lab sectors. This has happened directly within the ADAMACH project or other through research projects hosted at University of Trento. Fellows have been hired at top-universities in their home country, and one of them hired by the University of Trento.