Safe Adaptive Importance Sampling
This work addresses the problem of speeding up large-scale optimization algorithms for machine learning practitioners, offering an incremental improvement by integrating a novel sampling scheme into existing methods.
The paper tackles the computational infeasibility of adaptive importance sampling in optimization by proposing an efficient approximation using safe gradient bounds, resulting in significant speed-ups for coordinate-descent and stochastic gradient descent algorithms as verified by extensive testing.
Importance sampling has become an indispensable strategy to speed up optimization algorithms for large-scale applications. Improved adaptive variants - using importance values defined by the complete gradient information which changes during optimization - enjoy favorable theoretical properties, but are typically computationally infeasible. In this paper we propose an efficient approximation of gradient-based sampling, which is based on safe bounds on the gradient. The proposed sampling distribution is (i) provably the best sampling with respect to the given bounds, (ii) always better than uniform sampling and fixed importance sampling and (iii) can efficiently be computed - in many applications at negligible extra cost. The proposed sampling scheme is generic and can easily be integrated into existing algorithms. In particular, we show that coordinate-descent (CD) and stochastic gradient descent (SGD) can enjoy significant a speed-up under the novel scheme. The proven efficiency of the proposed sampling is verified by extensive numerical testing.