Jensen's inequality says that the average value of a function is biased toward the value where the derivative is closer to 0. That's where moving away from a sa

Jensen's inequality says that the average value of a function is biased toward the value where the derivative is closer to 0. That's where moving away from a sample point has the least impact on the output. It applies in scenarios that are "convex", which means that the derivative is monotonically increasing or decreasing, so "closer to 0" is a consistent direction.