The 3 magic words every materials scientist wants to hear

TLDR: Try out research, you will learn so much about yourself not to mention so much about one topic. However, the skills you learn can apply to anything.

GitHub - itszrong/2021-Faraday-Institution-Summer-Internship-ML-Battery-Manufacturing
Contribute to itszrong/2021-Faraday-Institution-Summer-Internship-ML-Battery-Manufacturing development by creating an account on GitHub.

My time with the Faraday Institution and the University of Warwick is coming to a close and I have some reflections.

I have realised I really enjoy doing research, getting into the details and trying to leverage ideas from other papers to make an improvement to my code. My research has been in data mining and machine learning for battery manufacturing. In materials science, the big idea is to explore process-structure-property (PSP) linkages.

the 3 magic words every materials scientist wants to hear

Starting the internship, I read many papers in machine learning and computational materials science to get up to speed on the current research landscape. There were not many that had been trying to numerically predict the capacity of electrodes, which is the property part of PSP, from the structural features: weight, thickness and porosity.

The machine learning pipeline I used was:

My work was in a mix of MATLAB and Python. I am hoping to turn my code into an app with React.js so that others can use it. It was written in the scikit-learn library.

Another issue I came across was that it is much more difficult to get large datasets with many features in materials science experiments as extensive time is required to measure each feature. As such, many preprocessing techniques are required.    

In the end, compiling my findings into a poster was really helpful in providing clarity of thought. See the poster here!

Tips for any students starting a research internship:

  • - keep a research log, explaining your motivations behind decisions and any results you collected
  • - progress in research takes much longer than you expect, you have to just sit there and try as many things as you can. When you get tired or annoyed, brew yourself a cup of tea :)
  • - when you get stuck and this will inevitably happen, read a lot of papers! (even if you don't think they are relevant!) You never know where the one solution or technique you needed can come from.

Special thanks to my supervisor Dr. Mona Faraji-Niri for the help and support!

Mona Faraji Niri
Senior Research Fellow, Warwick Manufacturing Group, University of Warwick - Cited by 133 - Battery Managment Systems - EV Powertrain - Machine Learning for Battery Manufacturing - Stochastic Control

Happy researching!