Reinforcement Learning and multi-armed bandits for improved Brain Computer Interfaces
*** Description of the research ***
Brain Computer Interfaces (BCI) provide a direct communication channel from the brain to a computer, bypassing traditional interfaces such as keyboard or mouse, and also providing a feedback to the user, through a sensory modality (visual, auditory or haptic). A target application of BCI is to restore mobility or autonomy to severely disabled patients, but more generally BCI opens up many new opportunities for better understanding the brain at work, for enhancing Human Computer Interaction, and for developing new therapies for mental illnesses.
In BCI, new modes of perception and interaction come into play, and a new user must learn to operate a BCI, as infants learn to explore their sensorimotor system: central to BCI operation are the notions of feedback and reward.
The goal of the research is to study the co-adaptation between a user and a BCI system in the course of training and operation. BCI will be considered under a joint perspective: the user’s and the system’s. From the user’s brain activity, features must be extracted, and translated into commands to drive the BCI system.
>From the point of view of the system, it is important to devise adaptive learning strategies, because the brain activity is user dependent and not stable in time. How to adapt the features in the course of BCI operation is a difficult and important topic of research. Reinforcement Learning (RL) and multi-armed bandit theory will be considered in order to address this question.
The postdoc will be funded by the french ANR CoAdapt project. The aim of CoAdapt is to propose new directions for BCI design, by modeling explicitly the co-adaptation taking place between the user and the system. See the description of the project and the consortium there:
*** Supervision ***
The research will be conducted under the supervision of Remi Munos and Emmanuel Dauce:
– Rémi Munos (http://chercheurs.lille.inria.fr/~munos/) is a senior INRIA researcher in the project team SEQUEL (Sequential Learning) of INRIA Lille (http://sequel.lille.inria.fr/).
– Emmanuel Daucé (http://emmanuel.dauce.free.fr/) is an associate professor at Centrale Marseille.
*** Background ***
We welcome applicants with a PhD degree in statistics / machine learning, or related fields and with interests in BCI, possibily with background in reinforcement learning or bandit theory.
*** Additional informations ***
The position is research only and is for one year, with possibility of being extended.
The starting date could be from April to December 2012.
To apply please send a CV to firstname.lastname@example.org or email@example.com