I hope I came to the right place to ask this question.

Back when I was at collage I studied machine and deep learning in-depth. My whole programme was based on those areas. I knew all underlying maths, even today I know how to derive backpropagation for any feed-forward network. Well, maybe I would need to take a peek. But I still understand the math and I can follow without problems. Back then (2017) I was even doing something with and researching GANs which were novelty back then.

Basically, I was at the hotspot at that time. I was working with all sorts of algorithms, from logistic regression, SVMs, to MLPs, CNNs, RNNs (mostly LSTM) and was trying already mentioned GANs. Oh, heuristics too: GAs, tabu search, simulated annealing, etc. There was also some NLP involved too.

And then after college I went to gaming industry, heh. I was/am still, during that time, working with ML/DL (and some OpenCV) but mostly easy, toy projects (although one project was real life project, but it was easy, I had to extract written digits from paper and classify them).

So, my question is, what is the best (and fastest) way to get back on the track considering my, I would say, pretty strong background? I saw that Kaggle has some courses on their site, for example course on DL is estimated 4 hours, course on feature engineering is also 4 hours. That is not a lot of time, but I am afraid it would be too easy for me and consequently a waste of time.

What are some good resources to refresh/relearn ML/DL considering my background?

  • 2
    $\begingroup$ Sounds to me like you don't need to relearn anything. You already know the fundamentals of ML, plus some pretty advanced stuff in DL. I would start by reading some of the landmark DL/ML papers from the last couple years, and brush up on the latest changes to Tensorflow and PyTorch. Then find an interesting vision or language challenge on Kaggle, and give it a try! $\endgroup$
    – zachdj
    Jan 14, 2020 at 18:10
  • 2
    $\begingroup$ +1 to apply your knowledge on Kaggle! You are in luck because cool NLP comps are active right now. $\endgroup$
    – Aditya
    Jan 14, 2020 at 18:19

1 Answer 1


If you enjoy reading, that's probably the best ML/DL book ever.

enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.