Recurrent neural networks having a probabilistic state component and state machines extracted from the recurrent neural networks
AssigneeNEC Corporation
InventorsCheng Wang, Mathias Niepert
Life of this patent
Estimating the term…
Abstract
A computer-implemented method includes instantiating a neural network including a recurrent cell. The recurrent cell includes a probabilistic state component. The method further includes training the neural network with a sequence of data. In an embodiment, the method includes extracting a deterministic finite automaton from the trained recurrent neural network and classifying a sequence with the extracted automaton.
Claims15
A computer-implemented method comprising:
- instantiating a neural network comprising a recurrent cell, the recurrent cell comprising a probabilistic state component;
- training the neural network with a sequence of data to generate a trained recurrent neural network;
- and
- extracting a deterministic finite automaton from the trained recurrent neural network,
- wherein the extracting of the deterministic finite automaton from the trained recurrent neural network comprises directly reading centroids of the trained recurrent neural network without learning from a clustering of a number of observed hidden states of the trained recurrent neural network.
A processing system comprising one or more processors configured to:
- instantiate a neural network comprising a recurrent cell, the recurrent cell comprising a probabilistic state component;
- train the neural network with a sequence of data to generate a trained recurrent neural network;
- and
- extract a deterministic finite automaton from the trained recurrent neural network,
- wherein the extracting of the deterministic finite automaton from the trained recurrent neural network comprises directly reading centroids of the trained recurrent neural network without learning from a clustering of a number of observed hidden states of the trained recurrent neural network.
A non-transitory computer-readable medium comprising code for causing one or more processors of a processing system to:
- instantiate a neural network comprising a recurrent cell, the recurrent cell comprising a probabilistic state component;
- train the neural network with a sequence of data to generate a trained recurrent neural network;
- and
- extract a deterministic finite automaton from the trained recurrent neural network,
- wherein the extracting of the deterministic finite automaton from the trained recurrent neural network comprises directly reading centroids of the trained recurrent neural network without learning from a clustering of a number of observed hidden states of the trained recurrent neural network.
Dependencies are inferred from the claim text.