Accelerated Stochastic Optimization Methods under Quasar-convexity

Understand

Non-convex optimization plays a key role in a growing number of machine learning applications.

• This motivates the identification of specialized structure that enables sharper theoretical analysis.
• One such identified structure is quasar-convexity, a non-convex generalization of convexity that subsumes convex functions.
• Existing algorithms for minimizing quasar-convex functions in the stochastic setting have either high complexity or slow convergence, which prompts us to derive a new class of stochastic methods for optimizing smooth quasar-convex functions.