Machine Learning tech

The previous tutorials Machine learning and Neural Networks give an intuition about what ML looks like and how it works under the hood. Now we will turn to topics about what a professional engineer deals with in his actual works:

Generalization Errors

The end goal of ML is to make the trained model perform well on new cases (in production). Thus we need to find ways to reduce the generalization error: Bias, Variance and Irreducible errors.

Challenges of Machine Learning

Most works of ML project are to resolve the challenges: bad data, bad algorithms, performance, and so on.

ML Workflow

The whole process of ML includes data preparation, cross-validation, fune-tuning the model, and so on.

Pack these steps in one pipeline and run the pipeline with one command.

Algorithms

Having a good understanding of how algorithms work can help you quickly home in appropriate model, the right training algorithm to use, and a good set of hyperparams for the task. Furthermore, the engineers can make some improvement of the algorithms for specific tasks.