Introduction to ML Strategy

Machine Learning projects, and especially Deep Learning projects, have a wide range of parameters and hyperparameters that must be tuned, as well as a wide range of techniques that can be applied (as discussed in earlier chapters) to improve the performance and optimize the learning problem.

However, it is not feasible to try every technique for every project. One must be able to identify issues, if any, and then try to use a specific set of techniques to rectify the issues. This process of having a specific set of solutions for specific problems is called orthogonalization.

It is, therefore, important to strategize the ML project in order to save time and still achieve the required results.