AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Inverse Modeling Problem in Machine Learning
Inverse modelling problem is starting from observed, sparse data and figuring out what the right barometers are for your simulation that is correct. In machine learning, you are using back propagation to find the sensitivity of your lost function with respect to the weight that youare i to train. We actually have been using differential programming methods, automatic differentiation in the fortrun world for almost two decades now. Wrightr: "automatic differentiation gives us a form using the chain role"