How can the knowledge of mathematical theory in machine learning be implemented in practical projects?

< H2 > background < / H2 >

I am a beginner in machine learning and complete a target recognition project at the request of the company. I learned a lot of mathematical knowledge in the early stage.

I understand that the role of mathematical knowledge in the field of machine learning is:

1. Build your own neural network structure: but now we all use structures like yolo,ssd, is it necessary to build our own structures?

2. Adjust parameters: we can adjust parameters according to various parameters during training (such as loss value, etc.), but how to combine with these mathematical theories?

Tensorflow artificial-intelligence machine-learning

Mar.25,2022

1, depending on your actual needs, if yolo,ssd does not meet some of your needs, you may need to transfer your study. So, whether you build it yourself, and how much you build depends on whether you can solve your problem. Of course, some projects may have special requirements and cannot use certain things, which is not up to you.
2, do more projects and you will know.

Previous: The style is out of order in Apple browser (why the problem is not the first time and the second time)

Next: The height of the innermost element is defined as a percentage, and the height is determined by the outermost element.