Neural network data entry system
First Claim
1. A data entry system comprising:
- a user interface which receives a sequence of one or more context text items input by a user;
processing circuitry configured to execute instructions to implement a plurality of software components, the software components comprising;
a predictor component trained to predict a next item in the sequence;
the predictor component comprising a plurality of text item embeddings stored in a vocabulary, each text item embedding representing a text item in a numerical form, and each text item embedding indicating weights for analysis of a representation of the text item in a trained neural network language model, with the text item embeddings having a plurality of different lengths;
a projection component which obtains text item embeddings for the context text items input by the user, the obtained text item embeddings being of the plurality of different lengths, and projects these to be of the same length;
the predictor component comprising a trained neural network which is fed the projected text item embeddings, having the same length, for use in the trained neural network language model, which computes a numerical output associated with the predicted next item.
2 Assignments
0 Petitions
Accused Products
Abstract
A data entry system is described which has a user interface which receives a sequence of one or more context text items input by a user. The data entry system has a predictor trained to predict a next item in the sequence. The predictor comprises a plurality of learnt text item embeddings each text item embedding representing a text item in a numerical form, the text item embeddings having a plurality of different lengths. A projection component obtains text item embeddings of the context text items and projects these to be of the same length. The predictor comprises a trained neural network which is fed the projected text item embeddings and which computes a numerical output associated with the predicted next item.
26 Citations
20 Claims
-
1. A data entry system comprising:
-
a user interface which receives a sequence of one or more context text items input by a user; processing circuitry configured to execute instructions to implement a plurality of software components, the software components comprising; a predictor component trained to predict a next item in the sequence; the predictor component comprising a plurality of text item embeddings stored in a vocabulary, each text item embedding representing a text item in a numerical form, and each text item embedding indicating weights for analysis of a representation of the text item in a trained neural network language model, with the text item embeddings having a plurality of different lengths; a projection component which obtains text item embeddings for the context text items input by the user, the obtained text item embeddings being of the plurality of different lengths, and projects these to be of the same length; the predictor component comprising a trained neural network which is fed the projected text item embeddings, having the same length, for use in the trained neural network language model, which computes a numerical output associated with the predicted next item. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)
-
-
17. A computer-implemented method comprising:
-
receiving a sequence of one or more context text items input by a user; accessing, from a memory, a plurality of text item embeddings stored in a vocabulary, each text item embedding representing a text item in a numerical form, and each text item embedding indicating weights for analysis of a representation of the text item in a trained neural network language model, with the text item embeddings having a plurality of different lengths; retrieving, from the memory, text item embeddings for the context text items input by the user, the retrieved text item embeddings being of the plurality of different lengths, and projecting the retrieved text item embeddings to be of the same length; and inputting the projected text item embeddings, having the same length, for use in the trained neural network language model, which computes a numerical output associated with a predicted next item of the sequence. - View Dependent Claims (18, 19)
-
-
20. One or more device-readable media with device-executable instructions that, when executed by a computing system, direct the computing system to perform operations comprising:
-
receiving a sequence of one or more context text items input by a user; accessing, from a memory a plurality of text item embeddings each text item embedding stored in a vocabulary, representing a text item in a numerical form, and each text item embedding indicating weights for analysis of a representation of the text item in a trained neural network language model, with the text item embeddings having a plurality of different lengths; retrieving, from the memory, text item embeddings for the context text items input by the user, the retrieved text item embeddings being of the plurality of different lengths, and projecting the retrieved text item embeddings to be of the same length; and inputting the projected text item embeddings, having the same length, for use in the trained neural network language model, which computes a numerical output associated with a predicted next item of the sequence.
-
Specification