Engineering topics
Zeng, Zheng
Publications and source records attributed to Zeng, Zheng.
Pseudogradient Training For A Class Of Neural Networks
Developmental second-order recurrent neural networks of special type modified to enhance stability in face of inputs beyond range of inputs on which trained. Second-order recurrent neural networks contain product feedback units and can be trained, by use of example inputs and outputs, to act as finite-state automatons. Particular second-order recurrent neural networks in question learn grammars in sense they are trained to generate binary responses to input training sequences of ones and zeros, each sequence being marked "legal" or "illegal" according to grammar to be learned.
Discrete Recurrent Neural Networks as Pushdown Automata
In this paper, we describe a new discrete recurrent model with discrete external stacks for learning context-free grammars (or pushdown automata).
Discrete Recurrent Neural Networks for Grammatical Inference
Recurrent neural networks have recently been shown to have the ability to learn regular and context-free grammars from examples. We show that while conventional analog recurrent networks try to form clusters in activation space to represent discrete states of the grammars during learning, and can be successful in doing so, the clusters so formed tend to become unstable as longer and longer test input strings are presented to the network.