Abstract :
Error-correcting codes are used in a variety of areas from computers to communications. Ideally, one simply looks at a received message which may contain errors, and decodes it into the error-free message. Unfortunately, this decoding process can be quite complicated and might not exploit the maximum error correction capabilities of the code. While for the simple Hamming code the decoding is direct, for more complicated codes decoding may be an NP-complete problem. Further the optimal decoding depends on the statistical properties of the error source, which might be very complex. For these reasons we are proposing the use of neural networks as decoders. In this paper, we report some experimental results which show that neural networks can be used to practically solve the general decoding problem mentioned above
