Learning Mappings with Neural Network

The authors extended the idea of training multiple tasks simultaneously on a partially shared feed forward network. A shared input subvector was added to represented common inputs of all tasks. A "cross training" mechanism was used to specify corresponding components between the two tasks. By examining output of one network in response to stimulus from the other network, we can test if the network can learn the correspondence which was not cross-trained. Two kinds of studies on binary tree mapping were conducted. The shared input was used to represent "relation" between tree nodes. The result shows the network can fill in the missing correspondence with sufficient training data.

• Type: Journal
• Year: 2009
• Where: ICAI
• Authors: Yefei Peng, Paul W. Munro
• Keywords: Artificial Intelligence, Feed Forward Network, ICAI 2009, Shared Input, Shared Input Subvector