Activation Function	The function that defines whether a neuron in a neural network is active.
Bayesian model	Bayes theorem is used with prior probabilities of past events for prediction.
CNN	Convolutional neural networks are a special form of artificial neural networks, strong when feature geometry is important as in images or spectral data.
Cross validation	Data is divided into folds, where every fold is used as a test set and average metrics across the folds are used to evaluate model statistics.
Feature	Observed variable used as input to the model for prediction.
Hyperparameter	Also known as metaparameters and used for tuning of the model training.
Latent variables	Features derived by mathematical transformation of features.
Overfitting	The model performs well on the training data but poorly on unknown data. Overfitting increases with variables and nonlinearity of the statistical model. Cross validation identifies overfitting.