An electromyographic signal (EMG) is formed by a component that is medically useful which contains information about the muscular activity, and an artifact, which is caused by external effects, e.g., Functional Electrical Stimulation (FES) signals. Several methods have been used to suppress such artifacts based on adaptive schemes. This paper presents a comparison between an adaptive filter using the Least Mean Squares algorithm (LMS) and a Multilayer Perceptron (MLP) using error Backpropagation (BP), Recursive Least Squares (RLS) and Iterative Least Squares (ILS) algorithms. The criteria for comparison are artifact suppression and speed of convergence.