[Seminar] "Action Selection & Reinforcement Learning in animals & robots" Dr. Benoît GIRARD

Dear all,

Neural Computation Unit (Doya Unit) would like to invite you to seminars as follows.

Speaker: Dr. Benoit Girard, ISIR, UPMC/CNRS
http://www.isir.upmc.fr/index.php?op=view_profil&id=109&old=N&lang=en

He has worked on many topics including rat-like robots, reinforcement learning, eye movement control, and basal ganglia modeling. He will present to us what he is up to in three talks:
Monday, December 5, 15:30-17:30
Tuesday, December 6, 10:00-12:00
Wednesday, December 7, 10:00-12:00
*Each will be one to two hours depending on how many questions and discussions there will be.

<First talk>
Title: Action Selection & Reinforcement Learning in animals & robots

Abstract: The basal ganglia are an essential component of the brain for action
selection, decision-making and reinforcement learning functions. I will
first present a few contributions dealing with the design of basal
ganglia models at the neural population level (Girard et al. 2003, 2008;
Lienard & Girard, 2014), from a neuro-robotics point of view (How can we
use or evaluate such models on robots? What are their dynamical
properties?) as well as from the neuroscience point of view (what are
the computations performed by this circuit?).
I will then briefly present some results dealing with reinforcement
learning: How can we integrate actor-critic algorithms of TD-learning
with models of the basal ganglia circuitry (Khamassi et al., 2005)? How
can we explain the behavior of rats in risk-varying environments
(Cinotti et al., in preparation)?
Finally, if we have enough time and if the audience is still alive, I
may present some results dealing with the interaction of perceptual
decision-making and biomechanics (Marcos et al., 2015).
In the next talks (Tuesday & Wednesday), I will present results also
mainly dealing with action selection and reinforcement learning, but in
the specific contexts of saccadic eye movements and of spatial navigation.

Cinotti, F., Fresno, V., Aklil, N., Coutureau, E., Girard, B., Marchand,
A., Khamassi, M. (in preparation). Dopamine blockade affects learning
rate and dynamic regulation of exploration.
Girard, B., Cuzin, V., Guillot, A., Gurney, K. N., & Prescott, T. J.
(2003). A basal ganglia inspired model of action selection evaluated in
a robotic survival task. Journal of integrative neuroscience, 2(02),
179-200.
Girard, B., Tabareau, N., Pham, Q. C., Berthoz, A., & Slotine, J. J.
(2008). Where neuroscience and dynamic system theory meet autonomous
robotics: a contracting basal ganglia model for action selection. Neural
Networks, 21(4), 628-641.
Khamassi, M., Lachèze, L., Girard, B., Berthoz, A., & Guillot, A.
(2005). Actor–Critic models of reinforcement learning in the basal
ganglia: from natural to artificial rats. Adaptive Behavior, 13(2), 131-148.
Liénard, J., & Girard, B. (2014). A biologically constrained model of
the whole basal ganglia addressing the paradoxes of connections and
selection. Journal of computational neuroscience, 36(3), 445-468.
Marcos, E., Cos, I., Girard, B., & Verschure, P. F. (2015). Motor cost
influences perceptual decisions. PloS one, 10(12), e0144841.