Table of contents
-
Reminder about the gaussian distribution and introduction to decision making. Exercises on sequential ratio probability test (SPRT) that brings the notion of likelihood and ends up with a process of decision making. Possibility to add slightly more complex models using urgency-gating etc... In the case of drift-diffusion model, we try to estimate a non-varying binary random variable.
-
This second session introduces the notion of prior (ie. we have some knowledge about the system we're observing) and how to mix them with likelihood in order to get an optimal posterior estimation. In this session, we will introduce Bayes' theorem and optimal combination of prior and likelihood on a toy example.
-
This third session will be about continuous time varying random variable which is not fully observable (for example, assessing the position of your limb while performing a movement). We will use an optimal combination of prior and likelihood (aka, Kalman filter) as before except that in this case, the RV can take an infinity of possible values.
In order to run the different tutorials, please click on the following links
| Run | View | |
|---|---|---|
| Tutorial 1 |  |
 |
| Tutorial 2 | |
|
| Tutorial 3 | |
|
| Tutorial 4 | |
|
The goal of the projects will be to analyse and model real life data collected during experiments. More details to come