In this work we studied at a practical level how computation with action potentials (spikes) can be performed. We address the problem of programming of a dynamical system modeled as a neural network and considering both, hardware and software implementations. For this, we considered a discrete-time spiking neuron model, which was introduced by Soula et al. in 2006, and called BMS. On one hand, we proposed an efficient method to properly estimate the parameters (delayed synaptic weights) of a neural network from the observation of its spiking dynamics. Thus, the idea was to avoid the underlying NP-complete problem (when both weights and inter-neural transmission delays are considered in the parameters estimation). So far, our method defines a Linear Programming (LP) system to perform the parameters estimation. Another aspect considered in this work was the fact that we included a reservoir computing mechanism (hidden network) as to increase the computational power as to add robustness. Furthermore, these ideas are applied to implement input-output transformation for learning the implicit parameters of the corresponding transfer function.