Some Approximation Properties of Projection Pursuit Learning Networks

摘要: This paper will address an important question in machine learning: What kind of network architectures work better on what problems? A projection pursuit learning has a very similar structure to one hidden layer sigmoidal neural network. general method based continuous version regression is developed show that works angular smooth functions than Laplacian functions. There exists ridge function approximation scheme avoid the curse dimensionality for approximating L2(θd).