TY - JOUR
TI - LOSSGRAD: Automatic Learning Rate in Gradient Descent
AU - Wójcik, Bartosz
AU - Maziarka, Łukasz
AU - Tabor, Jacek
TI - LOSSGRAD: Automatic Learning Rate in Gradient Descent
AB - 
	In this paper, we propose a simple, fast and easy to implement algorithm LOSSGRAD (locally optimal step-size in gradient descent), which automatically modifies the step-size in gradient descent during neural networks training. Given a function f, a point x, and the gradient ▽xf of f, we aim to find the step-size h which is (locally) optimal, i.e. satisfies:

	h = arg min f(x - t▽xf).
	            t≥0

	 

	Making use of quadratic approximation, we show that the algorithm satisfies the above assumption. We experimentally show that our method is insensitive to the choice of initial learning rate while achieving results comparable to other methods.
VL - 2018
IS - Volume 27
PY - 2018
SN - 1732-3916
C1 - 2083-8476
SP - 47
EP - 57
DO - 10.4467/20838476SI.18.004.10409
UR - https://ejournals.eu/en/journal/schedae-informaticae/article/lossgrad-automatic-learning-rate-in-gradient-descent
KW - gradient descent
KW - optimization methods
KW - adaptive step size
KW - dynamic learning rate
KW - neural networks