Math major who has an interest in machine learning

[John Ralph]

I'm a second year student and I've been researching about machine learning. I'd like to start learning more, and I'm wondering if someone could advise me on the appropriate modules to take which would help me in this goal. List of modules: http://www.ucl.ac.uk/maths/courses/undergraduates/

For year 2 I have taken:
Analysis 3
Algebra 3
Mathematical Methods 3
Fluid Mechanics

and I have a choice of four modules, I have chose:
Analysis 4
Algebra 4
Probability and statistics
Computational Methods

I do not know a lot of programming so I am hoping with self learning and the computational methods course I will learn more. I am asking if my choice of four modules is appropriate, and what should I look at choosing for years 3 and 4?

 

[berkeman]

I'm a second year student and I've been researching about machine learning. I'd like to start learning more, and I'm wondering if someone could advise me on the appropriate modules to take which would help me in this goal. List of modules: http://www.ucl.ac.uk/maths/courses/undergraduates/

For year 2 I have taken:
Analysis 3
Algebra 3
Mathematical Methods 3
Fluid Mechanics

and I have a choice of four modules, I have chose:
Analysis 4
Algebra 4
Probability and statistics
Computational Methods

I do not know a lot of programming so I am hoping with self learning and the computational methods course I will learn more. I am asking if my choice of four modules is appropriate, and what should I look at choosing for years 3 and 4?

What is "Machine Learning"? Sorry, I'm not familiar with that term.

 

[atyy]

What is "Machine Learning"? Sorry, I'm not familiar with that term.

Machine learning involves solving problems like speech recognition, or getting machines to explore an environment to find an object. Speech recognition involves fitting hidden markov models, so can be seen as a branch of statistics, while the exploration task is sometimes approached via reinforcement learning, so it can be seen as a branch of control theory. Here is a recent paper about getting the machine to play Atari (an old task): http://arxiv.org/abs/1312.5602. Some standard texts are

https://www.amazon.com/dp/0387310738/?tag=pfamazon01-20
Pattern Recognition and Machine Learning
Christopher Bishop

https://www.amazon.com/dp/0262018020/?tag=pfamazon01-20
Machine Learning: A Probabilistic Perspective
Kevin Murphy

 

[John Ralph]

Anyone have any advice?

 

[MarneMath]

They look like good courses for a future study into machine learning. For future courses, just focus more on statistics, simulations, and numerical analysis. Statistics because it's hard to get into ML without knowing something about testing. Simulations because it's a good introduction on how to apply statistical techniques. Numerical analysis because it's important to understand convergences and the different problems that arise with algorithms.

 

[John Ralph]

They look like good courses for a future study into machine learning. For future courses, just focus more on statistics, simulations, and numerical analysis. Statistics because it's hard to get into ML without knowing something about testing. Simulations because it's a good introduction on how to apply statistical techniques. Numerical analysis because it's important to understand convergences and the different problems that arise with algorithms.

Would the pure mathematics modules be of any use also? I'm very interested in analysis and would like to study it further in my degree.

 

[MarneMath]

Naturally, Analysis is useful, if for no other reason is that is a course that teaches a lot of the same techniques for proofs you'll need a higher level maths. Plus, in most advance statistics courses, you'll view probability in terms of a measure, so it's helpful to understand a measure. As for the Algebra 4, I never really had a use for rings or groups in my work. However, it's my personal opinion that it's interesting and that's often a good enough reason to study anything.

 

[John Ralph]

Naturally, Analysis is useful, if for no other reason is that is a course that teaches a lot of the same techniques for proofs you'll need a higher level maths. Plus, in most advance statistics courses, you'll view probability in terms of a measure, so it's helpful to understand a measure. As for the Algebra 4, I never really had a use for rings or groups in my work. However, it's my personal opinion that it's interesting and that's often a good enough reason to study anything.

Thank you for the reply, may I ask - what type of work do you do?

 

[MarneMath]

I work for a telecommunication company making predictive models on customer behaviors. It sounds uninteresting on the surface, but the intersect of mathematics and computer science is a happy spot for me. I've also done work about predictive texting. That was fun :).