4 Comments
Aug 31, 2021Liked by Andrey Kurenkov, Hugh Zhang

Hello all! I'd really like to see some more articles about interpretability, explicability, saliency mapping etc - it's what I'm working on right now: (shameless self promotion: https://arxiv.org/abs/2103.05108) and I feel is very much of the zeitgeist at the moment. We're making more and more complex and powerful models - we need ways to understand and learn from them.

Expand full comment
author

Thanks for the recommendation!

Expand full comment

You should try to improve the audio quality more (getting rid of reverb, static noise, air conditioners). A number of online services or downloadable programs can help with this, and I don’t want to push any specific one.

Expand full comment
author

Thanks for the suggestion! We actually do so already using Adobe Audition. But the results vary depending on the nature of the background noise and quality of original audio etc. We'll aim to keep improving, of course.

Expand full comment