Basic Inequalities for First-Order Optimization with Applications to Statistical Risk Analysis
By: Seunghoon Paik , Kangjie Zhou , Matus Telgarsky and more
We introduce \textit{basic inequalities} for first-order iterative optimization algorithms, forming a simple and versatile framework that connects implicit and explicit regularization. While related inequalities appear in the literature, we isolate and highlight a specific form and develop it as a well-rounded tool for statistical analysis. Let $f$ denote the objective function to be optimized. Given a first-order iterative algorithm initialized at $θ_0$ with current iterate $θ_T$, the basic inequality upper bounds $f(θ_T)-f(z)$ for any reference point $z$ in terms of the accumulated step sizes and the distances between $θ_0$, $θ_T$, and $z$. The bound translates the number of iterations into an effective regularization coefficient in the loss function. We demonstrate this framework through analyses of training dynamics and prediction risk bounds. In addition to revisiting and refining known results on gradient descent, we provide new results for mirror descent with Bregman divergence projection, for generalized linear models trained by gradient descent and exponentiated gradient descent, and for randomized predictors. We illustrate and supplement these theoretical findings with experiments on generalized linear models.
Similar Papers
On the Gradient Complexity of Private Optimization with Private Oracles
Machine Learning (CS)
Makes private computer learning faster and more efficient.
Bounds on inequality with incomplete data
Econometrics
Measures wealth gaps even with messy data.
Progressively Sampled Equality-Constrained Optimization
Optimization and Control
Solves hard math problems faster by guessing less.