There are many research areas in machine learning that are well-suited for people with a strong background in math. Some of which are:

You can search and explore more areas and choose one best suited to you

An Empirical Study of ML Algorithms for Stock Daily Trading Strategy.

Well… obviously nothing is impossible, but the idea that you could just waltz in and download some deep learning package and with a few mouse clicks, start doing publishable research in ML or AI is a bit naive. And yet, I’m glad you asked this question since it’s worth dispelling such ideas from the get go.

Now, all exploration and research begins with experimentation. There was indeed a time in physics when random tinkering with simple apparatus might have yielded major insights. But you’re several hundred years too late for hoping to do world class physics by random tinkering.

Publishable research, by definition, means you are aware and understand what’s been published in the area you are working in . In short, you understand the ML literature.

Take a look at the most recent NeurIPS or ICML conferences. The level of math in these conferences is now well above what you would have learned as an undergrad in engineering or CS. It takes years just to build up enough math background to barely follow some of the papers.

Collected.

Ammar Mujahid

I have seen so many people who are talking about machine learing - machine learing all the day , but if you ask the maths and background of algorithm behind then seems quite few will explain that. Well in my point of view A machine Learning is a Pure mathematics the major difference is that ML engineer knows coding . If we Exclude that part then the is no difference . As an mathematical background if some one wants to work in the ML then he/she can do lot of things.

Go for the optiomisation problem. There are number of problems needs to be addressed.

Go for the development of new algorithm to build up the neural networks .

Develop a new and better solution for the machine learning and deep learning algorithms .

Works in the sub domains of Natural Language processing , reinforcement based lerning and complex problem of biology , chemistry and physics .

Definatly you will find your intrest in some of the domains if you want to work .

The future of mathematics is bright in the field of computer science . As CS is applied mathematics with coding background. @

Thanks.

There is various area for research in ML. However according to me following worthy areas for you :

1. Bayesian and Statistical Machine Learning : The core theory of machine learning is statistical and Bayesian computation, regardless of whether one is using deep learning or predictive analytics. We develop computational techniques with a statistical basis to handle complex knowledge structures, process large scale data, and harness available prior knowledge, demanded of modern artificial intelligence applications.

https://www.monash.edu/it/dsai/machine-learning/research-projects/bayesian-and-statistical-machine-learning

2. Time Series Analysis: Time series is now ubiquitous and their analysis is an exciting field of research. Applications we work on include: working out if an insect flying through a sensor is a vector of the Nile virus, understanding if an area imaged by satellite images is prone to fire, or activity recognition from the accelerometers in your smartphone.

https://www.monash.edu/it/dsai/machine-learning/research-projects/time-series-analysis

I am a student majoring in Mathematics and Applied Mathematics in BSc and Computational Mathematics in MSc. Currently, I am majoring in Bioinformatics using some machine learning and feature selection methods, which are based on sparse statistical learning models. Early, I read some papers by Prof. Robert Tibshirani. And then, I work on the biomarkers discovery project. It makes me see the practical application of mathematics. When we collaborate with doctors in hospitals, our results also make them interested.