Finally, we suggest a modification to weight-normalization, which improves its performanceonlarge-scaletasks. 2

There aretwointeresting andwell-studied computational paradigms aimed attackling uncertainty.

We show in this work that reinforcement learning can be successfully applied to decoding short to moderate length sparse graph-based channel codes.

The hope is that these predictions allow the algorithm to circumvent worst case lower bounds when the predictions are good, and approximately match them otherwise. The precise definitions and guarantees vary with different settings, but there have been significant successes in applying this framework for many different algorithmic problems, ranging from general online problems to classical graph algorithms (see Section 1.2 for a more detailed discussion of related work, and [35] for a survey). In all of these settings it turns out to be possible to define a "prediction" where the "quality" of the algorithm (competitive ratio, running time, etc.) depends the "error" of the prediction.