Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: How to gain practical machine learning skills?
20 points by LDGcSmefqM on Dec 31, 2018 | hide | past | favorite | 7 comments
TLDR: I know the theoretical side of things in ML but lack practical experience. How can I fix that?

I am a CS student graduating in about half a year. I have taken a few ML classes in school. All of them went deep into the math (i.e., I know how things work, mostly) but very light on the "practical tricks" (i.e., I don't know how/where to use them). As a result, I do not have much experience writing code for ML stuff (save for some very simple assignments). For e.g., I have not used things like TensorFlow and PyTorch before.

What is your advice for me to gain practical ML skills? My gut feelings is that I need to do some exercises (a la leetcode). Where can I find these exercises? What other resources (e.g., books, courses, blogs, etc) should I look into?

Thanks and happy new year. :)




Try some Kaggle challenges first. It will get you familiar with the tools and basic workflows.

Once you've done a couple of challenges, try to find a task of your own and attempt to apply ML to it. Ideally go all the way and collect your own dataset. This will give you experience in formulating your problem, defining what is a good/acceptable solution and how a dataset can be created to solve it.


disagree. Kaggle challenges are very hard for the one who has no (previous) practical skill.


I enjoyed Andrew Ng's class on Coursera, which introduces basic concepts. For "real world," maybe find a dataset on data.world and try to figure out how to answer a question that is interesting to you?

https://www.coursera.org/learn/machine-learning


I'm going through this one right now, and really enjoying it. Being able to learn at my own pace is great. The course is math heavy, but that's to be expected. I'm enjoying using Octave for the programming.


The TensorFlow tutorials might be a reasonable next step. Google created them to help working engineers learn the necessary skills to apply TensorFlow to Google's real world problems. So there's a meaningful intersection between the tutorials and "the real world."


I just started working through Google's free course. Tomorrow I'm posting about my day one experience here: https://kagia.github.io/2018/12/29/learning-machine-learning...

As it says on the linked post I'm not learning just for knowledge sake but I very much intend to apply what I learn.


Kaggle?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: