An instructional automatic chat method based on deep reinforcement learning
A kind of intensive learning and guiding technology, applied in special data processing applications, instruments, electrical digital data processing, etc., can solve problems such as lack of decision-making ability, perception problems and helplessness
Inactive Publication Date: 2019-02-26
BEIJING UNIV OF TECH
View PDF5 Cites 6 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0006] Deep learning has strong perception ability, but lacks certain decision-making ability; while reinforcement learning has decision-making ability, it is helpless for perception problems
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
[0029] The present invention will be described in further detail below in conjunction with specific embodiments and with reference to the accompanying drawings.
[0030] The flowchart of the method of the present invention is as figure 1 As shown, it specifically includes the following steps:
[0031] Step 1, text word segmentation preprocessing.
[0032] Step 1.1, read the text and perform word segmentation.
[0033] Segment the user's input text, for example, segment "what's the weather in Beijing today" into "today", "Beijing", "weather", and "how". And perform vectorized representation by category. Since there are few categories, one-hot encoding is used.
[0034] Step 1.2, read the segmented words into representations of category vectors.
[0035] Words after word segmentation are represented by entity categories, for example, "today", "Beijing", "weather", "how" are recognized as "time", "place", "weather", "question words".
[0036] Step 1.3, represent the segmente...
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 discloses an instructional automatic chat method based on depth reinforcement learning, which belongs to the field of oral conversation system. Making use of the powerful perception of deep learning and the superior decision-making ability of reinforcement learning, the teaching module can give demonstration decisions directly or give additional rewards to instruct the decision-making module to make the right strategy. Whether the mode selection control teaching module gives the demonstration decision directly or the evaluation of the decision module is determined. The decision-making module constantly adjusts its own decision-making situation according to the evaluation given by the teaching module and the evaluation of the user as the optimization goal. This method increases the generalization ability of the system, increases the learning ability of the system, so that the system has stronger adaptability. Under this framework, the oral dialogue system can be trained more effectively, and the answers can be safer, more reasonable and more orderly.
Description
technical field [0001] The invention relates to the technical field of natural language processing, in particular to a model method for continuously optimizing a dialogue system through interaction with an environment based on deep reinforcement learning. Background technique [0002] A chat system is a service agent that can communicate with humans through everyday spoken language. Such systems will play an increasingly important role in our interactions with technology. Chat systems have a wide range of applications, from voice-enabled mobile apps to car navigation assistants, smart homes, tutoring systems and (in the not-too-distant future) service robots that assist us in our daily tasks. Chat systems have broad application scenarios and market demands, so it is of great significance to study how to improve the naturalness, coherence, stability, and intelligence of dialogue systems. [0003] Traditional chat systems are built based on artificial templates or knowledge ...
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): G06F16/332
Inventor 贾熹滨史佳帅刘洋曾檬苏醒郭黎敏
Owner BEIJING UNIV OF TECH