Abstract: In an approach to localizing tree-based convolutional neural networks, a method includes creating a first tree-based convolution layer (TBCL) corresponding to a tree, where the tree includes a first plurality of nodes and a node that has been indicated to be a first pivotal node. The first TBCL includes a second plurality of nodes and a second pivotal node having a feature vector based on node data from the first pivotal node. The method also includes creating a second TBCL corresponding to the tree. The second TBCL may include a third plurality of nodes. The method further includes determining a feature vector a third pivotal node in the third plurality of nodes based on the feature vectors from: (i) the second pivotal node, (ii) a parent node of the second pivotal node, and (iii) a child node of the second pivotal node.

Abstract: Systems and methods for training a neural network model are disclosed. In the method, training data is obtained by a deep neural network (DNN) first, the deep neural network comprising at least one hidden layer. Then features of the training data are obtained from a specified hidden layer of the at least one hidden layer, the specified hidden layer being connected respectively to a supervised classification network for classification tasks and an autoencoder based reconstruction network for reconstruction tasks. And at last the DNN, the supervised classification network and the reconstruction network are trained as a whole based on the obtained features, the training being guided by the classification tasks and the reconstruction tasks.