Named entity identification method based on gradient neural network structure search
A technology for named entity recognition and network structure, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve problems such as complex neural network structures, and achieve the effect of reducing workload, training time, and speeding up
Active Publication Date: 2021-01-26
沈阳雅译网络技术有限公司
View PDF2 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Similar to tasks such as named entity recognition, the neural network structure itself is relatively complex, and it is difficult to apply structure search technology to this task
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0040] The present invention will be further elaborated below in conjunction with the accompanying drawings of the description.
[0041] The invention relates to a named entity recognition method based on gradient neural network structure search. The method builds a bidirectional long-short-term memory network word feature extractor; builds a bidirectional cyclic neural network as a sequence coding layer; builds an attention inference layer based on multi-head attention , using gradient-based structure search techniques to obtain the internal structure of recurrent neural units in recurrent neural networks.
[0042]The named entity recognition method based on gradient neural network structure search of the present invention comprises the following steps:
[0043] 1) Obtain common data sets for named entity recognition tasks through data websites, and process them into a conll format suitable for named entity recognition tasks;
[0044] 2) Use cyclic neural network and other p...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention relates to a named entity identification method based on gradient neural network structure search, which comprises the following steps: acquiring a common data set of named entity identification tasks through a data website, and processing the common data set into a conll format; obtaining distributed representations of words and characters in the data, and splicing the distributed representations to serve as model input; building a main body model of the named entity recognition task and a loop computing unit search structure; carrying out normalization processing; optimizing the operation weight matrix of the recurrent neural unit and the overall parameters of the model at the same time; converting the continuous structural representation into a discretized structure; building a complete model structure of a named entity recognition task, and performing training and parameter optimization; and performing named entity identification by using the complete model after training convergence, and representing model performance by using accuracy. According to the method, a gradient-based structure search method is applied to implementation of named entity recognition taskswith complex structures, and adjustment is performed according to the particularity of the named entity recognition tasks, so that the performance of the named entity recognition tasks is improved.
Description
technical field [0001] The invention relates to a named entity recognition technology, in particular to a named entity recognition method based on gradient neural network structure search. Background technique [0002] In recent years, with the widespread dissemination and use of deep learning technology, methods based on neural networks have achieved amazing success in many fields. The performance of neural network-based methods on specific tasks often depends on the structure of the neural network, so most of the researchers' energy is focused on designing a better network structure. With the continuous advancement of research in various fields, more and more excellent neural network structures have been proposed, and the neural network structures applied to various tasks have become more and more complex, which means that experiments relying on artificially designed neural network structures The error cost and time cost will become more unbearable, so structure search te...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06F40/295G06N3/04G06N3/08
CPCG06F40/295G06N3/08G06N3/044G06N3/045Y02D10/00
Inventor 杜权
Owner 沈阳雅译网络技术有限公司