A0865
Title: Theory of deep convolutional neural networks
Authors: Dingxuan Zhou - University of Sydney (Australia) [presenting]
Abstract: Deep learning based on deep neural networks with network architectures has been powerful in practical applications but is less understood theoretically. The network structures such as convolutional architectures give essential difficulty, making the theory different from the classical one for fully connected neural networks. A mathematical theory of approximating and learning functions or operators is presented by deep convolutional neural networks and related schemes.
