Physics from Machine Learning’s perspective

Mriganka Nath
3 min readJul 16, 2019

If we see the starting of Science, it was with the people trying to understand their surroundings. This thirst of curiosity had done miracles to the ‘human’ revolution. With some theories dated back to the pre-Christ era starting with the Heliocentric theory, then came Galileo with his ideas and telescopes. After that Newton ruled the scene with his actions and the reaction gave Einstein the idea for relativity and with Bohr and Schrödinger down came Quantum Mechanics. But the scene of Physics is slowly deteriorating, very few new theories are coming and giving Experimental Physics a clear field to grow on.

But can this problem be solved in some way, can the observations and the data that was collected through the experiments can be used someway? The first thought when a large dataset is available and we want to find some relationship in it, Machine Learning comes into the picture.

Machine Learning is a subset of the Artificial Intelligence field, where we teach machines to learn. And how machines learn? by observing and training on data, a lot of processed data.

Machine Learning starts where general coding ends.

The most recent application of Machine Learning in Physics is the photo of the Blackhole. Many people are believing that it’s just a picture taken from the telescope, but it’s not. The image of the Blackhole is a result of an algorithm processed on the huge data sets taken by the telescopes all around the globe. This algorithm, called CHIRP which stands for ‘ Continuous High-resolution Image Reconstruction using Patch priors’ showed machine learning’s power and how wide it can be used, something which would have otherwise not been possible.

The scope of Machine Learning is enormous and this is just the beginning. And this field requires more people to contribute. New ideas and algorithms are constantly making our lives easy. This doesn’t stop here, there are few areas where ML is not impacting. Physics is still waiting to find it’s lost glory, the field is wide open and waiting to be solved. But the revolution has started and maybe soon, we can see the impact and how ML solves the mysteries of Physics.

For more understanding of how the picture of blackhole was taken, these links might come useful (the first video is highly recommended)

If someone is interested in Machine Learning, you can take this great MOOC by Andrew Ng



Mriganka Nath

high dimensions go brrrrr; I work with Neural Networks;