Posts
A comprehensive list of resources to learn advanced mathematics for machine learning by Aryansh Omray in Towards Data Science.
Mathematics
forms the basis of most of the machine learning algorithms. Therefore,
it is imperative to have a good grasp of mathematics to understand
machine learning. While most of the data scientists are aware of basic
mathematical concepts such as Linear Algebra, Statistics, etc. but many
of them are not aware of some deep mathematical concepts that can help
them have a clearer understanding of how an algorithm works or allow
them to understand the latest research in machine learning.
Photo: Kaboompics .com from Pexels
In this article, I have shared resources for advanced mathematics courses, which help machine learning. The topics discussed in this article are Convex and Non-Convex Optimization, Information Theory, Probabilistic Graphical Models, etc.
The list of resources is given so that it assumes the reader’s familiarity with basic concepts such as Linear Algebra, Probability Theory, Multivariable Calculus, and Multivariate Statistics. It is vital to understand these essential topics to understand the material presented in the advanced courses present in this article...
The resources in this article can be used to start a Ph.D. degree, where a thorough understanding of mathematical concepts relating to the research topic is expected from a student.
The plan is mainly divided into the following parts:-
- Convex Optimization
- Probabilistic Graphical Models
- Non-Convex Optimizations
- Information Theory
The list is never-ending, but the following four topics I have discussed here are essential in machine learning and highly transferrable to other engineering fields...
What to do next?
After completing the courses mentioned above, you can try to learn more courses in mathematics related to machine learning. For example, Measure Theory, Tensor Algebra, Mathematical Modelling, etc. are such other topics.
Other than the courses, you can now pick any theoretical machine learning paper presented at a top conference such as NIPS, ICML, ICLR, etc. and read and try to reproduce the results of the paper. The learning from this plan can also be used to start your research at any company or start a Ph.D. degree.
Source: Medium
