Thanks for asking! @rachel (mainly) and I (a bit) have been working on her numerical linear algebra course, which she is now halfway through teaching. She is recording it so we’ll be putting it online later. It’s extremely cool - we’ve found a bunch of recent advances that aren’t well appreciated but are a very big deal. And Rachel has figured out how to explain what is normally a very mathematically intensive subject in a classic fast.ai code-first top-down way!
Before I start working on the lung cancer dataset I wanted to try to make sure I was at or close to the current state of the art on computer vision, so for the last week I’ve been working on the Planet Kaggle competition. I’ve already learnt a lot. I’m working on it with @brendan (we’ll combine our teams later in the competition, but for now there is plenty of friendly rivalry!)
I’m advising www.doc.ai and www.platform.ai , both of whom are looking for good deep learning practitioners BTW so any alum of part 2 who are interested (and can be in the SF bay area) feel free to ask me for an intro).
I’m interested in what @sravya8 and team are doing at https://www.impactai.org/ and have been looking into how fast.ai can best support “AI For Good” projects. It’s still early days but there are some interesting possibilities.
Right now, the things that would help us the most are:
- Trying to replicate or beat the best results from the Kaggle Data Science Bowl lung cancer competition, and making your code and annotations available as open source
- Finding any parts of the part 2 notebooks that aren’t well described, and writing up clear and readable notes that we could include in the notebooks (ideally I’d like them to be able to be reasonably standalone)
- Telling your stories (on your blog etc) of your projects and how you’re using the course in your work, hobbies, education, etc.