public abstract class AbstractDifferentiableFunction<T>
extends java.lang.Object
This provides a separation between the functions and optimizers, that lets us test optimizers more effectively by generating convex functions that are solvable in closed form, then checking the optimizer arrives at the same solution.
Constructor and Description |
---|
AbstractDifferentiableFunction() |
Modifier and Type | Method and Description |
---|---|
abstract double |
getSummaryForInstance(T dataPoint,
ConcatVector weights,
ConcatVector gradient)
Gets a summary of the function of a singe data instance at a single point
|
public abstract double getSummaryForInstance(T dataPoint, ConcatVector weights, ConcatVector gradient)
dataPoint
- the data point we want a summary forweights
- the weights to usegradient
- the gradient to use, will be updated by accumulating the gradient from this instance