A Simplified Natural Gradient Learning Algorithm

Topmoumoute Online Natural Gradient Algorithm

Natural gradient is a gradient descent technique which uses the inverse of the covariance matrix of the gradient. Using the centrallimit theorem, we prove that it yields the direction that minimizes the probability of overfitting. However, its prohibitive computational cost makes it impractical for online training. Here, we present a new online version of the natural gradient which we coin TONG...

Natural Gradient Deep Q-learning

This paper presents findings for training a Q-learning reinforcement learning agent using natural gradient techniques. We compare the original deep Q-network (DQN) algorithm to its natural gradient counterpart (NGDQN), measuring NGDQN and DQN performance on classic controls environments without target networks. We find that NGDQN performs favorably relative to DQN, converging to significantly b...

Online gradient descent learning algorithm†

This paper considers the least-square online gradient descent algorithm in a reproducing kernel Hilbert space (RKHS) without an explicit regularization term. We present a novel capacity independent approach to derive error bounds and convergence results for this algorithm. The essential element in our analysis is the interplay between the generalization error and a weighted cumulative error whi...

Gradient echoes: simplified.

Natural Gradient Works Eciently in Learning

When a parameter space has a certain underlying structure, the ordinary gradient of a function does not represent its steepest direction but the natural gradient does. Information geometry is used for calculating the natural gradients in the parameter space of perceptrons, the space of matrices (for blind source separation) and the space of linear dynamical systems (for blind source deconvoluti...