Which of the following statements about parametrized predictors are true? Select all that apply.

A parametrized predictor is any model that depends on a set of tunable parameters, such as a neural network or a linear regression. These models are trained by adjusting the parameters so that the predictions match labeled data, usually using gradient‑based optimization. Because the number of parameters can be large, a parametrized predictor can fit the training data very well but may also overfit if not regularized. In practice, they can represent both linear and highly nonlinear relationships, and they can be regularized to improve generalization. Thus, key truths are that they are adjustable, trainable with data, can overfit, and can be regularized.