artificial intelligence - Does a pulsed (spiking) neural network outperform a recurrent neural net? -

Both pulsed and recurrent neural networks can model timing information models. But I'm not sure which model is better than the computational cost, does it pay to use more complex brain neural networks or will it require very little calculation with recurrent nerve outgoing work? Does the pulsed net gather faster?

Thanks

I do not believe you are seeing the problem properly way. The most important asset of a neural network (or any predictive model) is the accuracy, if I am much more accurate (and predictive), then I will spend 10 times more time in the model.

There are several standard techniques for assessing the estimated power of your model, such as * * leave one out cross validation # * leave-many-out cross verification * * Fisher Randomization (http: // en .wikipedia.org / wiki / Ronald_Fisher)

There are also several guides to create principles for projective models, such as * * Okram Razor * * Overfitting Avoid (http: //web.engr .oregonstate.edu / ~ tgd / classes / 534 / slides / part10.pdf)
* Penalties for overfitting (http://en.wikipedia.org/wiki/Regularization_ (mathematics))

More information Here are some places to Kne

Bottom line: Go for the simplest model that can explain your data