Artificial Neural Network Discussion

Hii..

Basically my project is a Brain controlled wheel chair and currently I am testing my algorithms for pre recorded data sets I have downloaded from BCI competition-3. Basically what I did was first I used the filters (band pass filter (0.5-100hz) and Notch filter (centred at 50Hz) ) and then, I've extracted the features using the Wavelet Packet Decomposition Method. The Feature vector I've got contains 17 features for each trail (each trail is represented by a row vector in the feature matrix) and I have a total of 268 trails, (135 for class-0 and 133 for class-1).Now, I want to implement ANN on that to classify my trails in class-0 and class-1.Basically, I trained the network with 133 trails of each class and then for the test trails, I carried out the features. Now, I am unable to get more than 50% accuracy for the test trails and I want to improve my results so if there is someone who can help me with this, kindly let me know. It's Urgent.

Please help.

I didn't find the right solution from the Internet.

References:-
sourceforge.net/p/fann/discussion/323464/thread/dd13498a/
Corporate AdvertisingCorporate Advertising

Thanks!