DeepAI AI Chat
Log In Sign Up

Margins, Shrinkage, and Boosting

by   Matus Telgarsky, et al.

This manuscript shows that AdaBoost and its immediate variants can produce approximate maximum margin classifiers simply by scaling step size choices with a fixed small constant. In this way, when the unscaled step size is an optimal choice, these results provide guarantees for Friedman's empirically successful "shrinkage" procedure for gradient boosting (Friedman, 2000). Guarantees are also provided for a variety of other step sizes, affirming the intuition that increasingly regularized line searches provide improved margin guarantees. The results hold for the exponential loss and similar losses, most notably the logistic loss.


page 1

page 2

page 3

page 4


Implicitly Maximizing Margins with the Hinge Loss

A new loss function is proposed for neural networks on classification ta...

A straightforward line search approach on the expected empirical loss for stochastic deep learning problems

A fundamental challenge in deep learning is that the optimal step sizes ...

AdaBoost and Forward Stagewise Regression are First-Order Convex Optimization Methods

Boosting methods are highly popular and effective supervised learning me...

Convergence of Gradient Descent on Separable Data

The implicit bias of gradient descent is not fully understood even in si...

Using a one dimensional parabolic model of the full-batch loss to estimate learning rates during training

A fundamental challenge in Deep Learning is to find optimal step sizes f...

Bandwidth-based Step-Sizes for Non-Convex Stochastic Optimization

Many popular learning-rate schedules for deep neural networks combine a ...