Hidden Markov Models: problem to solve for Master's thesis

[hcl14]

I apologize for asking this actually fuzzy question, but I really hope I can get some useful advice.

This year I'm finishing a Master's degree in the field of mathematics. I want to make some research and write a graduate thesis that (perhaps) could be interesting to my potential employer in the field of machine learning (data analytics, software development, etc) and will give me some really useful experience in "actual" tasks. My supervisor, after listening to me, let me propose my own topic in the field of Hidden Markov Models (Previously I was doing courseworks on Markov chains (coupling)).

So, currently I'm studying HMM applications in Natural Langauge Processing and Process Mining to find if I can do any research there and write my own project (I'm mathematician, so I need to perform some mathematical research or at least adapt an existing mathematical approach).

I'll be very grateful to everyone who will give me any advice or guidelines on the possible problem for me to solve in my thesis (HMM).

Thank you!

 

[mmwave]

Thank you for reaching out and sharing your situation with us. As a fellow scientist, I can definitely understand your desire to conduct meaningful research and gain valuable experience in your field.

First of all, congratulations on completing your Master's degree in mathematics! That is a great achievement and I'm sure it has equipped you with strong analytical skills that will be beneficial in your future research.

From what you have shared, it seems like you have a good understanding of Hidden Markov Models and their applications in Natural Language Processing and Process Mining. My advice would be to focus on a specific problem or application within these fields that interests you the most. This will not only make your research more engaging for you, but it will also make it more relevant to potential employers.

One possible approach could be to look into the limitations of current HMM models in these areas and see if you can come up with a new or improved method to address those limitations. This would involve conducting a literature review to identify gaps in the existing research and then proposing your own solution or approach.

Another idea could be to apply HMMs to a different field or problem that has not been explored before. This could potentially open up new avenues for research and contribute to the advancement of HMMs in general.

In terms of performing mathematical research, you could consider exploring the underlying principles and assumptions of HMMs and how they can be modified or extended to improve their performance.

Overall, my advice would be to choose a problem or application that you are passionate about and that aligns with your skills and interests. This will not only make your research more enjoyable, but it will also make it more impactful.

I wish you all the best in your research and I'm sure you will come up with a great thesis topic. Good luck!