Using BNN to Make a Prediction
Closed this issue · 2 comments
Hi, I have a question about using Blitz to make a prediction. Take the Boston dataste as an example, I can get the loss and accuracy for the model and I'm just wondering what can I do to make a prediction i.e I'm going to load the X_train again into the trained model to compare the prediction values to the true values.
Forgive my limit knowledge of neural networks since I had neber done anything in this field. : )
Best Regards
Hello. To perform the inference, you can either perform one feedforward pass on your model and interpret the prediction as one monte carlo sample or perform X feedforward pass, interpret your prediction as a random variable of which you are gathering the distribution via monte carlo and then check for variance and mean to quantify uncertainty.
Closing due to staleness.