Demos
 
Projects
Research Topics
Software
Seminars
Events
News
Members
Agenda
Demos
 
 
Items
Tracking the hidden state of a partially observable dynamical system with a simple recurrent self-organizing map : an experimental analysis reveals an unexpected richness of dynamics
 
  
> MALIS Home > Demos > Tracking the hidden state of a partially observable dynamical system with a simple recurrent self-organizing map : an experimental analysis reveals an unexpected richness of dynamics

MAchine Learning and Interactive Systems

Tracking the hidden state of a partially observable dynamical system with a simple recurrent self-organizing map : an experimental analysis reveals an unexpected richness of dynamics
 
  by Fix Jeremy
 
 

On this page, you will find some videos illustrating the behavior of the recurrent self-organizing map presented in the paper "Tracking the hidden state of a partially observable dynamical system with a simple recurrent self-organizing map : an experimental analysis reveals an unexpected richness of dynamics".

Sequence ABCDEFEDCB


Sequence AAAAAAAF


Sequence ABCDEFEDCB then Sequence ABCBAFEDEF


Sequence DEFE with noisy observations (gaussian with variance 0.05

Sequence ABCDEFEDCB with a perturbed state


We can also train the recurrent self-organzing map to represent the state of a POMDP. We consider a grid world from which only the x position is observed and therefore the observation is ambiguous. The self organizing map then provides an input to an adaptive controller with a linear representation of the Q-values, the features being RBF on the winning position of the self organizing map. The video below shows the performance of the algorithm as learning goes on.