A Unified Gradient-Descent/Clustering Architecture for Finite State Machine Induction

By Sreerupa Das, Michael C. Mozer. 1993

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

On the Binding Problem in Artificial Neural Networks (2020, background): “…inductive biases in neural networks that enable them to efficiently learn about symbols and the processes for manipulating them (examples of such an approach are abound, even in early connectionist research, e.g. Smolensky (1990); Pollack (1990); McMillan et al. (1992); Das and Mozer (1993)).”
Extraction of rules from discrete-time recurrent neural networks (1996, methodology): “, 1994, Giles et al., 1995b). Thus, it is natural to see how well recurrent networks learn what they can be proven to represent. This being said, it is not evident that recurrent networks are the best tools for learning regular grammars. However, regular language inference requires state information”
A Machine Learning Method for Extracting Symbolic Knowledge from Recurrent Neural Networks (2004, methodology): “The method proposed in Das and Mozer (1994) solves the problem of network instability, and thus simpler DFA extraction, by allowing discrete DFA representations to evolve during training.”
Rule Extraction from Recurrent Neural Networks: ATaxonomy and Review (2005, methodology): “This guaranteed the extraction of a deterministic machine since any state drift ( Das & Mozer 1994, Das & Mozer 1998) was avoided as the search was pruned when reentering already visited partitions.”
Analysis of Dynamical Recognizers (1997, background): “Others have been proposed since (Watrous & Kuhn, 1992, Frasconi et al., 1992, Zeng et al., 1994, Das & Mozer, 1994, Forcada & Carrasco, 1995) and a number of approaches to analysing recurrent networks have been developed.”