Special Session: Towards multimodal proactive interfaces using large-scale machine learning

The future multimodal interfaces which have several alternative input and output modalities and use context information extensively will face the problem of selecting the relevant data out of all the available sources. The proactive approach is to analyse all the explicit and implicit feedback of the user(s) and provide information that is likely to be the most interesting for the user. Efficient machine learning methods that can build practical models from large databases is here an important research topic as well as the development of representations suitable for the large quantities of data, such as images, audio, video and language and multimodal approaches. The scope of the special session includes (but is not limited to):

• efficient multimodal machine learning approaches
• novel applications to multimodal interfaces
• effective representations and modeling techniques for large-scale proactive interface tasks
• annotation techniques for multimodal data
• evaluation techniques for multimodal interfaces
• modeling methods for implicit relevance feedback in various modalities
• partial or delayed feedback
• contextual information retrieval methods
• multimodal augmented reality for displaying relevant information
• language and speech in conversational interfaces
• multimodal biometric interfaces
• modeling of user's state and interest
• adaptive multimodal interfaces
• multimodal mobile interfaces

Submissions

Please follow the instructions for the paper format and accessing the submission system.

Organisers:

Mikko Kurimo, Aalto University, Finland
Samy Bengio, Google Inc, CA, USA