Extracting Propositional Rules from Feed-forward Neural Networks — A New Decompositional Approach

3rd International Workshop on Neural-Symbolic Learning and Reasoning

Sebastian Bader ; Steffen Hölldobler ; Valentin Mayer-Eichberger

Others  :  http://CEUR-WS.org/Vol-230/04-bader.pdf
PID  :  12510

来源: CEUR

【 摘 要 】

In this paper, we present a new decompositional approach for the extraction of propositional rules from feed-forward neural networks of binary threshold units. After decomposing the network into single units, we show how to extract rules describing a unit's behavior. This is done using a suitable search tree which allows the pruning of the search space. Furthermore, we present some experimental results, showing a good average runtime behavior of the approach.