Natural language deep learning systems and methods
A deep learning and natural language technology, applied in the field of information processing, can solve problems such as insufficient training samples, obstacles to combining deep learning and natural language processing, and slow convergence of neural networks
Active Publication Date: 2020-12-15
FUJITSU LTD
View PDF8 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Simply initializing all unseen words to the same vector will cause the local convergence of the neural network to be too slow
[0005] In addition to the problem of dealing with word features, insufficient training samples are also a major obstacle to the combination of deep learning and natural language processing
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
[0021] Exemplary embodiments of the present invention will be described below with reference to the accompanying drawings. In the interest of clarity and conciseness, not all features of an actual implementation are described in this specification. It should be understood, however, that in developing any such practical embodiment, many implementation-specific decisions must be made in order to achieve the developer's specific goals, such as meeting those constraints related to the system and business, and those Restrictions may vary from implementation to implementation. Moreover, it should also be understood that development work, while potentially complex and time-consuming, would at least be a routine undertaking for those skilled in the art having the benefit of this disclosure.
[0022] Here, it should also be noted that, in order to avoid obscuring the present invention due to unnecessary details, only the device structure and / or processing steps closely related to the ...
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 natural language deep learning system and method. The system comprises an error calculating unit, when the natural language deep learning system is trained by the error calculating unit, error values of samples are calculated according to a loss function based on a sample pair, and the loss function is a combination of a similarity loss function and a category loss function, wherein the similarity loss function is defined base on the following rules that when real categories of the sample pair are the same, the difference between category prediction vector values of the sample pair should be small, while when the real categories of the sample pair are different, the difference between the category prediction vector values of the sample pair should be big; the category loss function is defined based on category errors of the sample pair. In the system, the loss function is designed based on the sample pair, so that the cost of loss studying based on the sample pair is reduced.
Description
technical field [0001] The present invention relates to the field of information processing, and more particularly to a natural language deep learning system and method. Background technique [0002] The combination of deep learning and natural language processing technology is a research hotspot in recent years. In the existing deep learning model system, using words as the basic feature unit is a common form of deep learning framework for natural language processing. Studies have shown that natural language processing features can effectively improve the learning performance of various tasks, so researchers usually use a variety of different word features to introduce more learnable factors. However, in actual operation, there will be the following two problems: [0003] 1. The natural language tools that generate word features all rely on word segmentation technology. Different word segmentation technologies lead to different word sequences in the same natural language ...
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 Patents(China)
IPC IPC(8): G06F40/284G06F40/30G06N3/08
CPCG06F40/211G06F40/284G06F40/40
Inventor 杨铭张姝孙俊
Owner FUJITSU LTD