Visualizing Deep Neural Nets - [Qure.ai]
Gosh and colleagues at Qure (2017) have published an insightful and illustrative post about visualizing deep learning networks. One of their goals is to build deep learning systems for the diagnosis of abnormalities in medical images. In their own words, the posts is a:
"a brief overview of the different perturbation based techniques for deep learning based classification models and their drawbacks." [source]
They are using chest X-rays for illustration purpose. I would say that to understand the write-up, one would have to be literate in machine learning methods as well as in medical diagnosis. So, unless you check on both of these requirements, this is not an easy read.
This is going to be a multi-part series; and, in the second part they are going to go into back-propagation visualization methods, which is as hard as it sounds :)
Anyway, you can see what it's all about by following the link below:
To stay in touch with me, follow @cristi
Cristi Vlad, Self-Experimenter and Author
This is actually some astounding work! Followed