Using FPGAs to Implement Artificial Neural Networks

In this paper we show and analyse the different alternatives used to implement, until now, artificial neural networks in FPGAs. At the moment, this is a very active research field, and still, there is a long way to travel. In this work, we focus on important aspects like: the neuron´s multiplier and activation function implementation, the storage and representation of the implicated data, the most habitual improvements and simplifications, the reconfigurable hardware systems used to implement artificial neural networks,... Ending this paper with the conclusions obtained from this analysis work. Among them, it is important to highlight that the use of FPGAs to implement ANN is not only feasible, but also presents a hopeful future.