Can Basic ML Techniques Illuminate Rateless Erasure Codes?

The recently developed rateless erasure codes are a near-optimal channel coding technique that guarantees low overhead and fast decoding. The underlying theory, and current implementations, of these codes assume that a network transmitter encodes according to a pre-specified probability distribution. In this report, we use basic Machine Learning techniques to try to understand what happens when this assumption is false. We train several classes of models using certain features that describe the empirical distribution realized at a network receiver, and we investigate whether these models can “learn” to predict whether a given encoding will require extra overhead. Our results are mixed.