Muhammad Kasim
Muhammad Kasim
Postdoctoral Research Assistant
Muhammad Firmansyah Kasim is a postdoctoral research assistant in the Atomic and Laser Physics, and a senior demonstrator in the Physics Computing lab. He obtained his DPhil from John Adams Institute in the University of Oxford. He got his BSc from Institut Teknologi Bandung (ITB) in Indonesia, with Electrical Engineering major. His main interest is to apply techniques from statistics, signal processing, computer science, and machine learning to solve problems in plasma physics.
He is a co-founder of Machine Discovery Ltd, a spin-out from the University of Oxford.
In his spare time, he blogs in mfkasim91.github.io.
I am a senior demonstrator in the Physics Computing lab for the first and second year of undergraduate in Oxford Physics department. Other than demonstrating, I am also responsible for the problem scripts revision to reach a higher standard.
I am also teaching math for first year physics students in Trinity College for MT 2019 & HT 2020.
I am very keen to learn numerical techniques from statistics, computer science, signal processing, and machine learning to solve problems in plasma physics. Some of the examples from my previous work are:
- 3D spectrometer. Using the concept of compressed sensing to obtain more information than the measured data in a spectrometer. It is a published patent with number WO/2019/025759.
- Maleo: integrating AI with simulations. A framework to easily integrate artificial intelligent (AI) algorithms and simulations without altering the simulations code.
- Inverting shadowgraphy or proton radiography. I used an optimal transport algorithm to retrieve the object's parameter from a shadowgram or a proton radiogram. Publication: https://arxiv.org/pdf/1607.04179.pdf
In my spare time, I do research in optimisation and machine learning. I have a peer-reviewed paper on optimisation that was accepted for the Bayesian Optimization workshop at NIPS 2016: M. F. Kasim and P. A. Norreys, Infinite dimensional optimistic optimisation with applications on physical systems, BayesOpt 2016 and arXiv:1611.05845 (2016).
I have another deep learning paper accepted for a workshop at NeurIPS 2019.
My fields of expertise for consultancy are optimization, Bayesian inference, and inverse problem.
I have helped a few companies to solve their inverse problems.