time step problem in neural networks

      Instead of normal array I tried using cells as input.
 input=[ c1 c2 c3 c4 c5 ];

where c1,c2,c3,c4 and c5 are the cells of the 3 different classes. (c1,c2-class one; c3-class two; c4,c5-class five) But when I gave it as input in the neural network pattern recognition window I got this error (bold)

Inputs 'input' is a 1x5 cell array of 57x12 matrices, representing dynamic data: 12 series of 5 timesteps of 57 elements.

Targets 'target' is a 3x5 matrix, representing static data: 5 samples of 3 elements.

Data selection have different number of time steps.

How to sort this out.