A Neural-Based Program Decompiler (2019)

&gt; <i>Empowered by the two-phase design, Coda achieves an average program accuracy of 82% on variousbenchmarks. While the Seq2Seq model with attention and the commercial decompilers yield 12%and 0% accuracy, respectively.</i><p>A commercial decompiler yeilding 0% accuracy sounds odd.

A curated list of awesome decompilation resources and projects.<p><a href="https:&#x2F;&#x2F;github.com&#x2F;nforest&#x2F;awesome-decompilation" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;nforest&#x2F;awesome-decompilation</a>

I&#x27;ve always wondered whether ml &#x2F; deep learning can be used efficiently for code deobfuscatuon

The model gives you approximate code based on a set of inputs and outputs you feed to it; each input and output helps the model shed light on a different execution path and improves the approximate code.<p>Still, not bad.

Do they have a release of this? Github or otherwise?

&quot;Neural-based&quot;, what a clickbaity way to say ml-assisted.