User Tools

Site Tools


analysis:cosmo2014:module4

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
analysis:cosmo2014:module4 [2014/08/16 16:02]
mvdm [Building the two-step decoder]
analysis:cosmo2014:module4 [2018/07/07 10:19] (current)
Line 13: Line 13:
 In the previous module, we applied the decoder to each time bin independently,​ using a flat spatial prior. In effect, this assumes that the place representation can move around arbitrarily from one time step to the next. Clearly, however, a rat cannot move around arbitrarily but instead moves subject to smoothness and continuity constraints! We can use this domain knowledge to improve the performance of our decoder. In the previous module, we applied the decoder to each time bin independently,​ using a flat spatial prior. In effect, this assumes that the place representation can move around arbitrarily from one time step to the next. Clearly, however, a rat cannot move around arbitrarily but instead moves subject to smoothness and continuity constraints! We can use this domain knowledge to improve the performance of our decoder.
  
-Our approach will be similar to Kalman filtering, in that we can construct a model of the rat's movement. We can then use this model to generate a prediction of the rat's position ($P(\hat{\mathbf{x}}_t)|P(\hat{\mathbf{x}}_{t-1}))$) which can function as a prior in out Bayesian decoder. The hat is meant to indicate that these are all estimates, since the decoder does not have access to the rat's true position $x$.+Our approach will be similar to Kalman filtering, in that we can construct a model of the rat's movement. We can then use this model to generate a prediction of the rat's position ($P(\hat{\mathbf{x}}_t)|P(\hat{\mathbf{x}}_{t-1}))$) which can function as a prior in our Bayesian decoder. The hat is meant to indicate that these are all estimates, since the decoder does not have access to the rat's true position $x$.
  
-Do do this, first we need to estimate $P(\hat{\mathbf{x}}_t)|P(\hat{\mathbf{x}}_{t-1}))$ from the data.+To do this, first we need to estimate $P(\hat{\mathbf{x}}_t)|P(\hat{\mathbf{x}}_{t-1}))$ from the data.
  
 ==== Estimating movement kernels ==== ==== Estimating movement kernels ====
Line 266: Line 266:
 Q = Q(:,​1:​end-1);​ Q = Q(:,​1:​end-1);​
  
-Q_tsd = tsd(Q_tvec_centers,​Q);​ 
-Q_tsd = restrict(Q_tsd,​run_start,​run_end);​ 
-</​code>​ 
- 
-Restricting to the times we are interested in: 
- 
-<code matlab> 
-%% only include data we care about (runs on the maze) 
 Q_tsd = tsd(Q_tvec_centers,​Q);​ Q_tsd = tsd(Q_tvec_centers,​Q);​
 Q_tsd = restrict(Q_tsd,​run_start,​run_end);​ Q_tsd = restrict(Q_tsd,​run_start,​run_end);​
analysis/cosmo2014/module4.1408219377.txt.gz ยท Last modified: 2018/07/07 10:19 (external edit)