CN110211573A - A kind of task-driven type dialogue decision-making technique based on neural network model - Google Patents
A kind of task-driven type dialogue decision-making technique based on neural network model Download PDFInfo
- Publication number
- CN110211573A CN110211573A CN201910450074.8A CN201910450074A CN110211573A CN 110211573 A CN110211573 A CN 110211573A CN 201910450074 A CN201910450074 A CN 201910450074A CN 110211573 A CN110211573 A CN 110211573A
- Authority
- CN
- China
- Prior art keywords
- model
- neural network
- task
- driven type
- confidence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 18
- 238000003062 neural network model Methods 0.000 title claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 16
- 230000004044 response Effects 0.000 claims abstract description 9
- 238000013527 convolutional neural network Methods 0.000 claims description 14
- 238000013528 artificial neural network Methods 0.000 claims description 13
- 230000000306 recurrent effect Effects 0.000 claims description 10
- 238000013461 design Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000004913 activation Effects 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 239000004576 sand Substances 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 abstract description 2
- 238000012546 transfer Methods 0.000 abstract description 2
- 230000003993 interaction Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000003780 insertion Methods 0.000 description 2
- 230000037431 insertion Effects 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/16—Speech classification or search using artificial neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- Acoustics & Sound (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Machine Translation (AREA)
Abstract
The present invention provides a kind of, and the task-driven type based on neural network model talks with decision-making technique, pass through the confidence state tracker and other partial parameters in training pattern, the coded vector generated by intention assessment model is as the confidence probability distribution over states for being intended to distribution and the generation of confidence state tracker, database is transferred to be inquired using the corpus being collected, use database search result, it is intended to distribution and probability distribution transfers to tactful combination of network to form system acting, it passes to and generates the response of combination of network output system, complete dialogue function, possess the features such as robustness is preferable, it is poor to solve model tormulation ability existing for task-driven type conversational system instantly, training difficulty is big, the limitation of model learnability is more, model training data volume is huge, model training reward mechanism is imperfect, practicability is poor etc. in specific field Problem.
Description
Technical field
The present invention relates to field of artificial intelligence more particularly to a kind of task-driven types pair based on neural network model
Talk about decision-making technique.
Background technique
In recent years, artificial intelligence technology application is gradually extensive, human-computer interaction smart home, intelligent medical treatment, public service,
There is no small application in the every field such as intelligent network connection automobile, and for the mankind for more convenient and fast life pursuit, having promoted can
Human language is interpreted, the autonomous robot for completing to respond or system are increasingly becoming a more active developing direction.Task-driven
Type takes turns one of the Main Morphology that dialogue is human-computer interaction current development more, takes turns dialogue system for task-driven type both at home and abroad at present more
The research of system has extensive development, primarily directed to the human-computer interaction of customer service realm, is mainly used in service trade.But
On vehicle, due to space is small, cost control requires height etc., human-computer interaction does not have cracking development, and emerging technology is not
It can be applied in the service to driver and passenger.The dialogue interaction of people's vehicle is related to a variety of proprietary knowledge in driving field, this makes
Instantly service class conversational system is unable to satisfy the needs of people-car interaction.
In the prior art, people-car interaction system mainly uses keyword, critical sentence matching way, in existing database
Find the answer of customer issue.The answer mode of mechanization in this way, seriously limits the expression way of driver and passenger, causes to talk with
Interaction excessively sequencing.And in the conversational system of single-wheel, the feelings such as unclear, faulty wording or colloquial style expression are expressed when the meaning of one's words occurs in user
When condition, the time for obtaining correct response can be greatly increased, and cause extreme influence to user experience.
For variety of problems existing for current technology and market, it is badly in need of a kind of for specific area, more wheels of specific crowd
Talk with the application of interaction technique, and big data is collected in recent years and remarkable break-throughs of the deep learning in natural language processing direction are
Realize that vehicle-mounted more wheel conversational system designs towards driver and passenger's demand provide new thinking.
Summary of the invention
The purpose of the present invention is to provide a kind of, and the task-driven type based on neural network model talks with decision-making technique, to mention
High driver and passenger are onboard for the quick convenience of the experience sense of onboard system and acquisition problem answers.
In order to achieve the above object, the present invention provides a kind of, and the task-driven type based on neural network model talks with decision
Method, comprising:
More wheels in one setting field of acquisition talk with training text, are based on LSTM network struction intention assessment model, and to institute
Intention assessment model is stated to be trained and cross validation;
Confidence state tracker is constructed to each information track, and builds convolutional neural networks and Recognition with Recurrent Neural Network more
New rule, with the training confidence state tracker;
The inquiry mode of design database and the database;
It is generated according to the output result three-phase matrixing of the figure identification model, confidence state tracker and database defeated
Unidirectional amount out, and the unidirectional amount of the output is adjusted based on language model;
It is trained using to wheel dialogue data, and all possible simultaneously according to the distribution of the global fiducial probability of dialogue state
Row dialog situation, to generate the dialogue movement of subsequent time proposition.
Optionally, include: based on the step of LSTM network struction intention assessment model
The each round dialogue in more wheel dialogues is encoded in being intended to identification model, obtains the coding of each round dialogue
Vector t;
LSTM network is constructed according to the following formula:
Wherein, ztFor the input for being t with sequential codingDistribution indicate, itIndicate input gate, ftIt indicates to lose
Forget door, otIndicate out gate, ci-1,ciIt indicates to recall state in short term.Wxc,WhcIt is trainable parameter, hi-1Indicate hidden layer.
Optionally, the confidence state tracker is followed with convolutional neural networks feature extractor with Jordan type by one
Ring neural network is constituted.
Optionally, the step for building the update rule of convolutional neural networks and Recognition with Recurrent Neural Network includes:
The convolutional neural networks are built, word intermediate features and sentence expression feature, and design feature vector are extractedFor the series connection of two convolutional neural networks derived characters, the input u of t-1 wheel is handled according to the following formulat, processing t-1 wheel
Response st-1:
According to the following formula with the both sides of sentence filling sentence before each convolution operation:
Wherein, vector ws, matrix Ws, bias term bsWith b 'sAnd scalar gφ,sIt is parameter,It is not mentioned when t wheel
And the probability of the value, the Recognition with Recurrent Neural Network weight of each value v are combined, and each activation primitive is being updatedWhen change it is special
Sign
Optionally, the inquiry content q of the databasetIt is realized by following formula:
Wherein, s ' is the information of input, SIIt is a group information slot,It is the output of confidence state tracker.
Optionally, according to the output z of the figure identification modelt, confidence state tracker outputAnd database is defeated
The truth vector x that result obtains outt, three-phase matrixing is carried out according to the following formula generates the unidirectional amount O of outputt:
Wherein, Wzo、Wpo、WxoIt is parameter,It is all series connection for summarizing confidence vector.
Task-driven type dialogue decision-making technique provided by the invention based on neural network model passes through in training pattern
Confidence state tracker and other partial parameters, the coded vector generated by intention assessment model is as intention distribution and confidence shape
The confidence probability distribution over states that state tracker generates is transferred to database to be inquired using the corpus being collected, uses data
Library searching result, intention distribution and probability distribution transfer to tactful combination of network to form system acting, pass to generation group of networks
Output system response is closed, dialogue function is completed, possesses the features such as robustness is preferable, solve task-driven type conversational system instantly
Existing model tormulation ability is poor, and training difficulty is big, and the limitation of model learnability is more, and model training data volume is huge, mould
The problems such as type training reward mechanism is imperfect, and practicability is poor in specific field.
Detailed description of the invention
Fig. 1 is that the present invention is based on the flow charts that the task-driven type of neural network model talks with decision-making technique;
Fig. 2 be the present invention towards vehicle-mounted more wheel dialogue overall framework figures;
Fig. 3 is recurrent neural network (RNN) confidence trace model of present invention binding convolutional neural networks (CNN).
Specific embodiment
A specific embodiment of the invention is described in more detail below in conjunction with schematic diagram.According to following description and
Claims, advantages and features of the invention will become apparent from.It should be noted that attached drawing is all made of very simplified form and
Using non-accurate ratio, only for the purpose of facilitating and clarifying the purpose of the embodiments of the invention.
As shown in Figure 1-Figure 3, a kind of task-driven type dialogue decision based on neural network model is present embodiments provided
Method includes the following steps:
Step 1: intention assessment includes following 3 sub-steps:
(1) the dialogue training text for acquiring driver and passenger's demand field, for training subsequent intention assessment model and setting
Believe state tracker;
(2) each round dialogue in more wheel dialogues is encoded in being intended to identification model, the coding talked with to
Measure t, ztThe input for being t with sequential codingDistribution indicate, construct LSTM network, by the output layer of final stepThe probability of different intentions is indicated as in driver and passenger's demand field, completes intention assessment modelling;Wherein,
ztFor the input for being t with sequential codingDistribution indicate, itIndicate input gate, ftIt indicates to forget door,
otIndicate out gate, ci-1,ciIt indicates to recall state in short term.Wxc,WhcIt is trainable parameter, hi-1Indicate hidden layer.
(3) for given intention assessment model training data set, intention assessment model is trained and cross validation,
Ultimately generate the analysis result of user's intention.
Step 2: confidence status tracking includes following 4 sub-steps:
(1) dedicated confidence state tracker is constructed to each information track, each confidence state tracker is had by one
Convolutional neural networks feature extractor and Jordan type Recognition with Recurrent Neural Network are constituted;
(2) every wheel Discourse Context background is modeled, feature vectorIt is special for the derivation of two convolutional neural networks
The user of the series connection of sign, processing t wheel inputs ut, the system response s of processing t-1 wheelt-1, calculation formula is as follows:
The dedicated convolution neural network computing of slot valueSentence expression is not only extracted, also extracts and removes Lexical label
Position, determine intermediate n-gram model insertion.In each dialogue, multiple matchings are such as observed, then to corresponding insertion
Summation.On the other hand, it if particular time-slot or value mismatch, is embedded in the n-gram of zero padding sky.In order to track vocabulary
The position of the label of change, with the both sides of sentence filling sentence before each convolution operation, the quantity of vector is by every layer of filtering
Device size determines.
(3) the update rule of design cycle neural network iterates to calculate the confidence state of content expressed by every wheel user
Probability;
Wherein, vector ws, matrix Ws, bias term bsWith b 'sAnd scalar gφ,sIt is parameter,It is not mentioned when t wheel
And the probability of the value, the Recognition with Recurrent Neural Network weight of each value v are combined, and each activation primitive is being updatedWhen change it is special
Sign
Confidence state tracker maintains multinomial point for each information track (can be used to the slot of Constrain Searching, such as stop)
Cloth p, and maintain each request slot (can challenge value slot, such as address) binary distributed.
(4) training confidence state tracker, passes through the output of confidence state tracker, data base querying content qtBy following
Algorithm is realized:
S ' is the information of input, SIIt is a group information slot,It is the output of confidence state tracker.The content is applied to
Database creates binary system true value vector x on the databaset, wherein 1 indicates that corresponding entity and inquiry are consistent, it can be deduced that
It is consistent with most probable confidence state.If x is endless when being all sky, associated entity pointer keeps mark random selection
A matching entities, entity pointer reference entity be used to form final system response.
Step 3: dialogue generates, and includes following 4 sub-steps:
(1) z for exporting intention assessment modelt, the output of confidence state trackerIt is obtained with database retrieval result
Truth vector xtInput, output indicate that O is unidirectionally measured in the output of system actingt, generate sentential form appropriate, information-based confidence
The individual probability of classification value in state is inessential, and be added together to be formed the summary confidence of each information track to
Amount.Indicated by three parts: total value probability, user indicate that they " are indifferent to " probability of this slot and do not mention slotting
The probability of slot.Finally, output is generated by three-dimensional matrixing:
Wherein, Wzo、Wpo、WxoIt is parameter,It is all series connection for summarizing confidence vector;
(2) generation system is adjusted according to the unidirectional amount of output in the sentence token that similar template is generated based on language model
Response;By from format surface list random sampling replace nonlinearized label, for example,<s.place>to place or
Lexical value is gone in the actual property value replacement in person area, the entity currently selected with database pointer, and token generating process is by one
Group pointer network enhancing, entity specific information is transferred in response.
(3) it is trained using more wheel dialogue datas;
(4) all possible parallel session situation is distributed according to the global fiducial probability of dialogue state, generates subsequent time
The dialogue of proposition acts.
The above is only a preferred embodiment of the present invention, does not play the role of any restrictions to the present invention.Belonging to any
Those skilled in the art, in the range of not departing from technical solution of the present invention, to the invention discloses technical solution and
Technology contents make the variation such as any type of equivalent replacement or modification, belong to the content without departing from technical solution of the present invention, still
Within belonging to the scope of protection of the present invention.
Claims (6)
1. a kind of task-driven type based on neural network model talks with decision-making technique characterized by comprising
More wheels in one setting field of acquisition talk with training text, are based on LSTM network struction intention assessment model, and to the meaning
Figure identification model is trained and cross validation;
Confidence state tracker is constructed to each information track, and builds the update rule of convolutional neural networks and Recognition with Recurrent Neural Network
Then, with the training confidence state tracker;
The inquiry mode of design database and the database;
It is single that output is generated according to the output result three-phase matrixing of the figure identification model, confidence state tracker and database
Vector, and the unidirectional amount of the output is adjusted based on language model;
It is trained using to wheel dialogue data, and all possible parallel right according to the distribution of the global fiducial probability of dialogue state
Situation is talked about, to generate the dialogue movement of subsequent time proposition.
2. the task-driven type based on neural network model talks with decision-making technique as described in claim 1, which is characterized in that base
Include: in the step of LSTM network struction intention assessment model
The each round dialogue in more wheel dialogues is encoded in being intended to identification model, obtains the coding vector of each round dialogue
t;
LSTM network is constructed according to the following formula:
Wherein, ztFor the input for being t with sequential codingDistribution indicate, itIndicate input gate, ftIt indicates to forget door,
otIndicate out gate, ci-1,ciIt indicates to recall state in short term.Wxc,WhcIt is trainable parameter, hi-1Indicate hidden layer.
3. the task-driven type based on neural network model talks with decision-making technique as claimed in claim 2, which is characterized in that institute
It states confidence state tracker and is constituted with convolutional neural networks feature extractor with Jordan type Recognition with Recurrent Neural Network by one.
4. the task-driven type based on neural network model talks with decision-making technique as claimed in claim 3, which is characterized in that take
The step for building the update rule of convolutional neural networks and Recognition with Recurrent Neural Network includes:
The convolutional neural networks are built, word intermediate features and sentence expression feature, and design feature vector are extractedIt is two
The series connection of a convolutional neural networks derived character handles the input u of t-1 wheel according to the following formulat, the response of processing t-1 wheel
st-1:
According to the following formula with the both sides of sentence filling sentence before each convolution operation:
Wherein, vector ws, matrix Ws, bias term bsWith b 'sAnd scalar gφ,sIt is parameter,It is not refer to this when t wheel
The Recognition with Recurrent Neural Network weight of the probability of value, each value v is combined, and each activation primitive is being updatedWhen change feature
5. the task-driven type based on neural network model talks with decision-making technique as claimed in claim 4, which is characterized in that institute
State the inquiry content q of databasetIt is realized by following formula:
Wherein, s ' is the information of input, SIIt is a group information slot,It is the output of confidence state tracker.
6. the task-driven type based on neural network model talks with decision-making technique as claimed in claim 4, which is characterized in that root
According to the output z of the figure identification modelt, confidence state tracker outputAnd the obtained true value of the output result of database to
Measure xt, three-phase matrixing is carried out according to the following formula generates the unidirectional amount O of outputt:
Wherein, Wzo、Wpo、WxoIt is parameter,It is all series connection for summarizing confidence vector.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910450074.8A CN110211573A (en) | 2019-05-28 | 2019-05-28 | A kind of task-driven type dialogue decision-making technique based on neural network model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910450074.8A CN110211573A (en) | 2019-05-28 | 2019-05-28 | A kind of task-driven type dialogue decision-making technique based on neural network model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110211573A true CN110211573A (en) | 2019-09-06 |
Family
ID=67788983
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910450074.8A Pending CN110211573A (en) | 2019-05-28 | 2019-05-28 | A kind of task-driven type dialogue decision-making technique based on neural network model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110211573A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110599065A (en) * | 2019-09-23 | 2019-12-20 | 合肥工业大学 | Pointer neural network-based multi-satellite emergency task planning method and system |
CN110619478A (en) * | 2019-09-23 | 2019-12-27 | 合肥工业大学 | Pointer neural network-based single-satellite emergency task planning method and system |
CN112328776A (en) * | 2021-01-04 | 2021-02-05 | 北京百度网讯科技有限公司 | Dialog generation method and device, electronic equipment and storage medium |
CN112579758A (en) * | 2020-12-25 | 2021-03-30 | 北京百度网讯科技有限公司 | Model training method, device, equipment, storage medium and program product |
WO2021129262A1 (en) * | 2019-12-26 | 2021-07-01 | 思必驰科技股份有限公司 | Server-side processing method and server for actively initiating conversation, and voice interaction system capable of actively initiating conversation |
CN113705652A (en) * | 2021-08-23 | 2021-11-26 | 西安交通大学 | Task type conversation state tracking system and method based on pointer generation network |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170053646A1 (en) * | 2015-08-17 | 2017-02-23 | Mitsubishi Electric Research Laboratories, Inc. | Method for using a Multi-Scale Recurrent Neural Network with Pretraining for Spoken Language Understanding Tasks |
CN108874782A (en) * | 2018-06-29 | 2018-11-23 | 北京寻领科技有限公司 | A kind of more wheel dialogue management methods of level attention LSTM and knowledge mapping |
CN109933659A (en) * | 2019-03-22 | 2019-06-25 | 重庆邮电大学 | A kind of vehicle-mounted more wheel dialogue methods towards trip field |
-
2019
- 2019-05-28 CN CN201910450074.8A patent/CN110211573A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170053646A1 (en) * | 2015-08-17 | 2017-02-23 | Mitsubishi Electric Research Laboratories, Inc. | Method for using a Multi-Scale Recurrent Neural Network with Pretraining for Spoken Language Understanding Tasks |
CN108874782A (en) * | 2018-06-29 | 2018-11-23 | 北京寻领科技有限公司 | A kind of more wheel dialogue management methods of level attention LSTM and knowledge mapping |
CN109933659A (en) * | 2019-03-22 | 2019-06-25 | 重庆邮电大学 | A kind of vehicle-mounted more wheel dialogue methods towards trip field |
Non-Patent Citations (2)
Title |
---|
TSUNG-HSIEN WEN ET AL.: "《A Network-based End-to-End Trainable Task-oriented Dialogue System》", 《PROCEEDINGS OF THE 15TH CONFERENCE OF THE EUROPEAN CHAPTER OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS》 * |
赵淑芳等: "《基于改进的LSTM深度神经网络语音识别研究》", 《郑州大学学报(工学版)》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110599065A (en) * | 2019-09-23 | 2019-12-20 | 合肥工业大学 | Pointer neural network-based multi-satellite emergency task planning method and system |
CN110619478A (en) * | 2019-09-23 | 2019-12-27 | 合肥工业大学 | Pointer neural network-based single-satellite emergency task planning method and system |
CN110619478B (en) * | 2019-09-23 | 2022-04-22 | 合肥工业大学 | Pointer neural network-based single-satellite emergency task planning method and system |
CN110599065B (en) * | 2019-09-23 | 2022-04-22 | 合肥工业大学 | Pointer neural network-based multi-satellite emergency task planning method and system |
WO2021129262A1 (en) * | 2019-12-26 | 2021-07-01 | 思必驰科技股份有限公司 | Server-side processing method and server for actively initiating conversation, and voice interaction system capable of actively initiating conversation |
CN112579758A (en) * | 2020-12-25 | 2021-03-30 | 北京百度网讯科技有限公司 | Model training method, device, equipment, storage medium and program product |
CN112328776A (en) * | 2021-01-04 | 2021-02-05 | 北京百度网讯科技有限公司 | Dialog generation method and device, electronic equipment and storage medium |
CN113705652A (en) * | 2021-08-23 | 2021-11-26 | 西安交通大学 | Task type conversation state tracking system and method based on pointer generation network |
CN113705652B (en) * | 2021-08-23 | 2024-05-28 | 西安交通大学 | Task type dialogue state tracking system and method based on pointer generation network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110211573A (en) | A kind of task-driven type dialogue decision-making technique based on neural network model | |
CN108874782B (en) | A kind of more wheel dialogue management methods of level attention LSTM and knowledge mapping | |
Chen et al. | Three-layer weighted fuzzy support vector regression for emotional intention understanding in human–robot interaction | |
Han et al. | Statistical‐based approach for driving style recognition using Bayesian probability with kernel density estimation | |
CN110427461B (en) | Intelligent question and answer information processing method, electronic equipment and computer readable storage medium | |
CN110046671A (en) | A kind of file classification method based on capsule network | |
CN107766876B (en) | Driving model training method, driver's recognition methods, device, equipment and medium | |
CN110619123A (en) | Machine reading understanding method | |
CN111008293A (en) | Visual question-answering method based on structured semantic representation | |
CN109241251A (en) | A kind of session interaction method | |
CN113010693A (en) | Intelligent knowledge graph question-answering method fusing pointer to generate network | |
CN107766787A (en) | Face character recognition methods, device, terminal and storage medium | |
CN112949647B (en) | Three-dimensional scene description method and device, electronic equipment and storage medium | |
CN106407874A (en) | Handwriting recognition method based on handwriting coordinate sequence | |
CN107526798A (en) | A kind of Entity recognition based on neutral net and standardization integrated processes and model | |
CN107662617A (en) | Vehicle-mounted interactive controlling algorithm based on deep learning | |
CN112464004A (en) | Multi-view depth generation image clustering method | |
CN105809201A (en) | Identification method and device for autonomously extracting image meaning concepts in biologically-inspired mode | |
CN106530341A (en) | Point registration algorithm capable of keeping local topology invariance | |
WO2021052137A1 (en) | Emotion vector generation method and apparatus | |
CN112115242A (en) | Intelligent customer service question-answering system based on naive Bayes classification algorithm | |
CN115187610A (en) | Neuron morphological analysis method and device based on graph neural network and storage medium | |
CN109344242A (en) | A kind of dialogue answering method, device, equipment and storage medium | |
Lin et al. | Intelligent decision support for new product development: a consumer-oriented approach | |
CN109933659A (en) | A kind of vehicle-mounted more wheel dialogue methods towards trip field |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190906 |