I am trying to improve my approach to Machine Learning and could use some practical advice.
I am particularly interested in best workflow and evaluation. I would like to compare approaches that people have actually used rather than just theoretical suggestions.
Are there any settings, tools, or workflow changes that made a noticeable difference?
One thing worth trying with How do you improve best workflow when working with Machine Learning? is a controlled comparison around workflow and improve. Use the same input for each test and note what changes in the output rather than judging from a single run. That way you can improve the workflow without losing track of what caused the improvement.
For How do you improve best workflow when working with Machine Learning?, I would separate the problem into setup, output quality and repeatability. workflow and improve fits into that because it can change the result quite a lot depending on the workflow. That way you can improve the workflow without losing track of what caused the improvement.
I think workflow and improve is the key part of this topic. A practical approach is to define the outcome you want, test it with a small example, and then check quality, speed and reliability separately. That way you can improve the workflow without losing track of what caused the improvement.