Machine Learning Compilers

[Amaroq_Starwind]

So, I've been doing some personal research regarding Machine Learning, and I came across some interesting stuff...

Blog: https://chriscummins.cc/2017/deep-learning-in-compilers/ 	<--	at the time of this writing, currrently a False Positive on MBAM web protection
Academic Paper: https://arxiv.org/pdf/1805.03441.pdf
Academic Paper: https://arxiv.org/pdf/1801.04405.pdf
Blog: http://jayconrod.com/posts/94/machine-learning-and-compiler-optimization
Quora Post: https://www.quora.com/What-changes-can-be-brought-in-compilers-if-we-include-machine-learning-deep-learning-algorithms

What applications could you think of for using Machine Learning in a compiler? I can easily imagine using Machine Learning to automatically debug code for compatibility issues, security vulnerabilities, and other miscellaneous bugs... it would be interesting to see what other things it could be used for.

[Amaroq_Starwind]

To amend this post, it's worth noting that not all types of Machine Learning are suitable for all applications, and many have their upsides and downsides. However, combining different types of Machine Learning could have its own benefits. Here is some further reading, for those who are new to Machine Learning. Even I learned quite a bit today!

https://towardsdatascience.com/types-of-machine-learning-algorithms-you-should-know-953a08248861
https://www.digitalvidya.com/blog/types-of-machine-learning/
https://en.wikipedia.org/wiki/Machine_learning

For those of you who are more deeply familiar with it however, especially the Malwarebytes team, definitely share your thoughts.