Get the latest tech news
Physics Informed Neural Networks
Welcome to my professional portfolio and blog!
The program covers a lot of topics, encompassing traditional machine learning methods (for example, timeseries forecasting which I recently discussed), fundamentals of computer vision models, all the way towards the latest and hottest generative AI techniques (which I hope I’ll get to talk about soon! With a PINN, you can use the autograd properties of your favorite tensor package to calculate the loss function at much greater accuracy, and in particular it doesn’t depend on the size of the mesh you use to train the model. This model has far-reaching importance throughout all of physics: at a simplified level, it describes systems such as sound waves in metals or electrical RLC circuits, while its quantum counterpart is at the root of the quantized energy description of atoms.
Or read this on Hacker News