Prev: how to use neural network to tell true or false of a realtime
Next: Miss something or Bug?
From: Walter Roberson on 21 Jul 2010 23:27
Simeon wrote:
> Thanks for the reply. The problem is: the new input from the subjects
> keeps coming in, there is no time to "Train it again". I am wondering
> whether there is a better way to do the two tasks simultaneously:
> 1. tell weather the current input is ture ("the most frequent feature
> vector" )
> 2. update the network

Do you have access to the distributed computing facilities?

If not, then how much of a load is running the experiment? I'm thinking
about the possibility of running a second matlab session on the same
machine, communicating with the first via UDP, with the machine running
the experiment sending the data via UDP to the second session which
would do the training and send the results back when polled (i.e., when
the machine running the experiment is in a state where it is convenient
to receive the new coefficients.)
 | 
Pages: 1
Prev: how to use neural network to tell true or false of a realtime
Next: Miss something or Bug?