Pith. sign in

Logistic Regression: Tight Bounds for Stochastic and Online Optimization

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
abstract

The logistic loss function is often advocated in machine learning and statistics as a smooth and strictly convex surrogate for the 0-1 loss. In this paper we investigate the question of whether these smoothness and convexity properties make the logistic loss preferable to other widely considered options such as the hinge loss. We show that in contrast to known asymptotic bounds, as long as the number of prediction/optimization iterations is sub exponential, the logistic loss provides no improvement over a generic non-smooth loss function such as the hinge loss. In particular we show that the convergence rate of stochastic logistic optimization is bounded from below by a polynomial in the diameter of the decision set and the number of prediction iterations, and provide a matching tight upper bound. This resolves the COLT open problem of McMahan and Streeter (2012).

fields

cs.GT 1

years

2025 1

verdicts

ACCEPT 1

representative citing papers

Prediction-Aware Learning in Multi-Agent Systems

cs.GT · 2025-01-31 · accept · novelty 6.0

A contextual optimistic multiplicative weights algorithm (POMWU) achieves static-game regret, equilibrium convergence, and social welfare guarantees in time-varying games when players can predict the changing state of nature with bounded error.

citing papers explorer

Showing 1 of 1 citing paper.

  • Prediction-Aware Learning in Multi-Agent Systems cs.GT · 2025-01-31 · accept · none · ref 41 · internal anchor

    A contextual optimistic multiplicative weights algorithm (POMWU) achieves static-game regret, equilibrium convergence, and social welfare guarantees in time-varying games when players can predict the changing state of nature with bounded error.