Normal Distribution and Gradient Descent
The Normal Distribution and Gradient Descent notebook shows how to estimate the mean and standard deviation of a normal distribution from data using maximal likelihood estimation (MLE). The MLE algorithm is implemented using gradient descent, which uses DiffKt to calculate the derivatives of the mean and standard deviation for use in the algorithm. In this example the data is stored in a tensor and the derivatives are calculated from a function over a tensor.