I'm a very technical person, but I never feel dumber than when doing mathematics. Self-teaching it is especially challenging. In many fields, one can learn to recognize general patterns and quickly apply them elsewhere. In math, I find that general pattern matching can be helpful for getting ideas, but there's so shortcut to working through the logic required in a rigorous mathematical proof. It's almost a different kind of thinking, and it takes a ton of effort.
If you're interested in self-teaching, I recommend an excellent Abstract Algebra book designed explicitly for that: https://today.williams.edu/books/abstract-algebra-a-student-.... This was the book that got me into rigorous mathematics and started me on the journey of doing a math PhD (defending this semester!). It only requires high-school algebra, and honestly, even that isn't really needed.
Several versions ago, there was a mod (maybe called "homeworld") that had a sink of you sending various resources through a portal back "home". It started simple with N iron plates, but then they wanted 10N oil barrels, and eventually red circuits. It also added some farming resources (which were affected by local pollution, so you had to plan where to put them). It was fun to have little goals, but I don't know that it's been updated recently.
I don't often self-promote, but my channel, DanDoesData, hosts the long-form type of series you may be looking for. Every video is a hour of live-stream where I research, design, and implement the model on the fly, explaining my thought process along the way.
One of my longest and most "real-world" projects was a (relatively simple) self-driving car based on single camera input. I used this model for a self-driving PowerWheels race. It later served as a starting point for a small autonomous RC car.
Here's the start of that playlist, but I've put together several projects. Working on a GAN for fake startup names right now.
Can you describe some of your methodology for detecting the fakes? Presumably you have trained a detector on a large data set; did you generate this yourself using several popular deepfake tools?
I was a little bummed to see no mention of `slimux`. It's a fantastic way to execute commands in a real shell (another tmux pane) from the comfort of your ViM session. By combining this with say IPython running in the shell, it's a powerful way to do data analysis. I wrote an ebook on this kind of workflow at http://dvbuntu.github.io/compute/
I really enjoy vim tabs for multiple files. Especially if I don't need to see all of them at once, but just quickly switch between them. `:tabnew foo.txt` to open another file, then `gt` to go to the next tab and `gT` to go back.
If you work interactively with data, you might also enjoy Slimux (https://github.com/epeli/slimux), a vim plugin that allows you to send lines from vim to an arbitrary tmux pane. I usually have IPython running in such a pane so I never have to copy+paste. I've got an ebook I've been kicking around describing this workflow in more detail here (http://dvbuntu.github.io/compute/posts/2050/01/01/workflow.h...).
I created a video course aimed at ML enthusiasts with software experience. It introduces Google's TensorFlow library, walks through code examples, and tries intuitively explain some of the statistics. https://www.packtpub.com/big-data-and-business-intelligence/...
But I also agree with other posters that you should study fundamental statistics first.
I actually studied exactly that! The optimal calculations for headway distance get a bit involved, but I coded up a modest algorithm and simple simulator. https://github.com/dvbuntu/oddities
Is it really "making it too easy" if you're applying your knowledge of the structure of the problem space to make it easier for the computer to solve? Certainly this isn't easy to do with every problem, but it seems like a better idea in general to start with parameters you suspect to be correct.
In the "swiss cake roll" the circular nature of the classes suggests using a sin or cos function, and the fact that they spiral out suggests also inputting magnitude information. Sure, you can just add more neurons that will end up computing the same thing, but we might as well give the computer a head start when we can.
Interesting concept. The aggregating of documentation and analysis together reminds me of a Jupyter notebook. If you add in a fully-featured CAD plugin and connect everything to AWS or whatever, does that approximate your product plan?
If you're interested in self-teaching, I recommend an excellent Abstract Algebra book designed explicitly for that: https://today.williams.edu/books/abstract-algebra-a-student-.... This was the book that got me into rigorous mathematics and started me on the journey of doing a math PhD (defending this semester!). It only requires high-school algebra, and honestly, even that isn't really needed.