We consider the problem of developing rapid, stable, and scalable stochastic gradient descent algorithms for optimisation of very large nonlinear systems. Based on earlier work by Orr et al. on adaptive momentum--an efficient yet extremely unstable stochastic gradient descent algorithm--we develop a stabilised adaptive momentum algorithm that is suitable for noisy nonlinear optimisation problems. The stability is improved by introducing a forgetting factor 0 ? ? ? 1 that smoothes the trajectory and enables adaptation in non-stationary environments. The scalability of the new algorithm follows from the fact that at each iteration the multiplication by the curvature matrix can be achieved in O (n) steps using automatic differentiation tools. We illustrate the behaviour of the new algorithm on two examples: a linear neuron with squared loss and highly correlated inputs, and a multilayer perceptron applied to the four regions benchmark task.
[1]
Sharad Singhal,et al.
Training Multilayer Perceptrons with the Extende Kalman Algorithm
,
1988,
NIPS.
[2]
Todd K. Leen,et al.
Optimal Stochastic Search and Adaptive Momentum
,
1993,
NIPS.
[3]
D. Marquardt.
An Algorithm for Least-Squares Estimation of Nonlinear Parameters
,
1963
.
[4]
Nicol N. Schraudolph,et al.
Fast Curvature Matrix-Vector Products for Second-Order Gradient Descent
,
2002,
Neural Computation.
[5]
Barak A. Pearlmutter.
Fast Exact Multiplication by the Hessian
,
1994,
Neural Computation.
[6]
Barak A. Pearlmutter,et al.
Automatic Learning Rate Maximization in Large Adaptive Machines
,
1992,
NIPS.