#### Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Options

# Neural network: el nino 3.4 anomalies index

Neural Network Delta Forecast for El Nino 3.4 Anomalies

A bit improvement on the error in comparison to SVR's 9% reduced to 7.9% for NN, but the max deviation range was increased a bit.

IMPORTANT: error for each of the 6 months was at the same level, not increasing as in the case of SVR

3 Layers, Input and Hidden each of 37 length, in other words the training samples vectors of length 37.

Output layer of length 6, for next 6 months forecasts.

300 sweeps of entire data, each step repeated 30 times, total of 9000 learning sessions.

Dara

• Options
1.

John

I assume N in NIPS is for Neural Networks. So I did this forecast for you to be on the safe side. Also I wanted to make sure the SVR results are not off.

Comment Source:John I assume N in NIPS is for Neural Networks. So I did this forecast for you to be on the safe side. Also I wanted to make sure the SVR results are not off.
• Options
2.

Code written in C, fully parallelized on 16 cpu server via Intel's OPENMP compiler technology.

Comment Source:Code written in C, fully parallelized on 16 cpu server via Intel's OPENMP compiler technology.
• Options
3.

Hello John

This algorithm assumes smooth functions and the manipulation of the error terms has calculus background to them.

Another version of this algorithm could be obtained for continuous functions which are not smooth, and no calculus smooth operators needed.

In this case the error term will be minimized by a global minimizer e.g. Differential Evolution and that could run might fast on the new GPU servers.

Dara

Comment Source:Hello John This algorithm assumes smooth functions and the manipulation of the error terms has calculus background to them. Another version of this algorithm could be obtained for continuous functions which are not smooth, and no calculus smooth operators needed. In this case the error term will be minimized by a global minimizer e.g. Differential Evolution and that could run might fast on the new GPU servers. Dara
• Options
4.

I will go thru the code this week and thoroughly test everything to make sure results are as accurate as possible

Comment Source:I will go thru the code this week and thoroughly test everything to make sure results are as accurate as possible