CN109615242A - A Recurrent Neural Network-Based and Cost-Sensitive Software Bug Assignment Method - Google Patents
A Recurrent Neural Network-Based and Cost-Sensitive Software Bug Assignment Method Download PDFInfo
- Publication number
- CN109615242A CN109615242A CN201811528909.9A CN201811528909A CN109615242A CN 109615242 A CN109615242 A CN 109615242A CN 201811528909 A CN201811528909 A CN 201811528909A CN 109615242 A CN109615242 A CN 109615242A
- Authority
- CN
- China
- Prior art keywords
- bug
- model
- csdbt
- neural network
- cost
- 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 25
- 230000000306 recurrent effect Effects 0.000 title claims description 11
- 230000001537 neural effect Effects 0.000 title 1
- 239000011159 matrix material Substances 0.000 claims abstract description 28
- 238000013528 artificial neural network Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 19
- 238000012360 testing method Methods 0.000 claims abstract description 15
- 230000003044 adaptive effect Effects 0.000 claims abstract description 8
- 238000012216 screening Methods 0.000 claims abstract description 8
- 230000000694 effects Effects 0.000 claims abstract 2
- 238000005457 optimization Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 4
- 230000035772 mutation Effects 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 claims description 4
- 230000002457 bidirectional effect Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 2
- 230000036461 convulsion Effects 0.000 claims 1
- 238000007781 pre-processing Methods 0.000 abstract 2
- 125000004122 cyclic group Chemical group 0.000 abstract 1
- 238000010200 validation analysis Methods 0.000 abstract 1
- 230000008569 process Effects 0.000 description 5
- 238000003745 diagnosis Methods 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000013522 software testing Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Artificial Intelligence (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Educational Administration (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Machine Translation (AREA)
Abstract
本发明公开了一种基于循环神经网络和代价敏感的软件bug分派方法，包括以下步骤：S1:对采集历史bug报告仓库中的原始数据集进行预处理；其中预处理包括筛选Bug报告、提取筛选出的Bug报告的文本信息、提取开发者活跃度信息；S2：在训练集上对CSDBT模型进行预训练；S3：使用自适应差分进化算法求解最优的误分类代价矩阵：验证集的数据进行求解，获取误分类代价矩阵的最优值；S4：将预训练好的的CSDBT模型和上一步得到的最优误分类代价矩阵相结合得到新CSDBT模型，最后输入测试集对新CSDBT模型进行测试。
The invention discloses a method for assigning software bugs based on cyclic neural network and cost-sensitive, comprising the following steps: S1: Preprocessing the original data set in the collection historical bug report warehouse; wherein the preprocessing includes screening bug reports, extracting and screening The text information of the bug report, and the developer activity information is extracted; S2: Pre-train the CSDBT model on the training set; S3: Use the adaptive differential evolution algorithm to solve the optimal misclassification cost matrix: the data of the validation set Solve to obtain the optimal value of the misclassification cost matrix; S4: Combine the pre-trained CSDBT model with the optimal misclassification cost matrix obtained in the previous step to obtain a new CSDBT model, and finally input the test set to test the new CSDBT model .
Description
Technical field
The present invention relates to software testing technology fields more particularly to a kind of soft based on Recognition with Recurrent Neural Network and cost-sensitive
Part bug allocating method.
Background technique
Software Bug is the software program problem in the process of running, their presence meeting is so that software failure, institute
It is the basis for guaranteeing software quality to repair software Bug in time.The warehouse software Bug is one kind for management software bug report
Efficient tool, the Bug report that user submits will store in the warehouse software Bug, and administrative staff are by reading these Bug report
To assign suitable developer to repair them.But being gradually expanded with software project scale, software bug and participation
The quantity for the developer that Bug is repaired all greatly increases, and consuming time is long, cost is efficient for the mode of this artificial carry out Bug appointment
Rate is low, and people need a kind of more efficient mode to complete the work of Bug appointment.So automation Bug dispatch technique is met the tendency of
And give birth to, Bug assignment problem is converted into text classification problem by machine learning by researchers, proposes numerous automatic Bug
Dispatch technique.But current many automatic Bug dispatch techniques only consider Bug report being dispatched to most suitable developer, without
The burden problem for considering developer is easy to cause a certain developer to overstock and the Bug repaired is largely needed to report, to drag slow project
Whole reparation process.
Summary of the invention
According to problem of the existing technology, the invention discloses a kind of soft based on Recognition with Recurrent Neural Network and cost-sensitive
Part bug allocating method, specifically includes the following steps:
S1: the raw data set in acquisition history bug report warehouse is pre-processed；Wherein pretreatment includes screening
Bug report, extracts developer's liveness information at the text information for extracting the Bug report filtered out；
S2: pre-training is carried out to CSDBT model on training set；
S3: optimal misclassification cost matrix is solved using adaptive differential evolution algorithm: the data for verifying collection are asked
Solution obtains the optimal value of misclassification cost matrix；
S4: the good CSDBT model of pre-training and optimal misclassification cost matrix obtained in the previous step are combined to obtain
New CSDBT model, recently enters test set and tests new CSDBT model.
Further, raw data set pre-process specifically in the following way:
S11: carrying out screening reserved state to bug report is the bug repaired, deletes the inefficient bug in vain and repaired；
S12: extracting text information from the bug report filtered out, chooses text information to be segmented, be gone at stop words
Reason deletes the excessive and very few word of frequency of occurrence；
S13: developer's liveness information is extracted, product and the identical developer of module information of current bug report are collected
Active sequences；
S14: pretreated data set is divided into training set, verifying collection and test set.
Further, in S2 it is specific in the following way:
S21: One-Hot encoding textual information and developer's information are used；
S22: the high-level characteristic of text information is extracted using bidirectional circulating neural network, is taken out using one-way circulation neural network
The high-level characteristic for taking liveness information merges two kinds of high-level characteristics in a manner of being multiplied between element；
S23: the conversion of high-level characteristic to developer's probability is completed；
S24: in training set data, to minimize cross entropy as target, optimization neural network model is extremely restrained.
Further, in S3 it is specific in the following way:
S31: initialization misclassification cost population wherein has individual, each individual represents a misclassification generation in population
The feasible solution of valence matrix；
S32: mutation operator and crossover operator based on adaptive generation generate a new misclassification cost population；
S33: calculating the fitness value of all individuals in population, by the misclassification cost matrix band respectively of each individual representative
Enter the good CSDBT model of pre-training, operation obtains correct rate score on verifying collection by the new CSDBT model of acquisition, this is correct
Rate score is the fitness value of a feasible solution；
S34: according to current individual with the fitness value of the corresponding individual of previous generation, therefrom selecting the higher individual of fitness,
Abandon the individual of performance difference；
S35: the mutation operator and crossover operator of adaptive adjustment population；
S36: it returns to S32 step and continues iteration, until reaching the maximum number of iterations of setting, and select last Dai Zhongshi
The highest individual of response is used as optimal solution.
Further, in S4 it is specific in the following way:
S41: on the CSDBT model for completing to instruct in advance, in conjunction with the misclassification cost matrix that optimization obtains, input is by pre- place
The test set of reason；
S42:CSDBT model will return to nominator's list that a length is K for each sample in test set.
By adopting the above-described technical solution, provided by the invention a kind of soft based on Recognition with Recurrent Neural Network and cost-sensitive
Part bug allocating method, this method uses differential evolution algorithm, is each data set under the premise of no any priori knowledge
The most suitable misclassification cost matrix of Optimization Solution.Based on original DeepTriage model, the order of text had both been taken into account
With the liveness information of developer, solve the problems, such as that data category is unbalanced also by misclassification cost matrix.Based on above-mentioned
Reason, present invention effectively prevents the same developers the overstocked problem of a large amount of bug reports occurs, accelerate the reparation of project into
Journey.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The some embodiments recorded in application, for those of ordinary skill in the art, without creative efforts,
It is also possible to obtain other drawings based on these drawings.
Fig. 1 is the flow chart of the method for the present invention
Fig. 2 is the flow chart of method part of detecting in the present invention.
Specific embodiment
To keep technical solution of the present invention and advantage clearer, with reference to the attached drawing in the embodiment of the present invention, to this
Technical solution in inventive embodiments carries out clear and complete description:
A kind of software bug allocating method (Cost based on Recognition with Recurrent Neural Network and cost-sensitive as depicted in figs. 1 and 2
Sensitive Deep Bug Triage, CSDBT), specifically includes the following steps:
S1: the raw data set in acquisition history bug report warehouse is pre-processed；Wherein pretreatment includes screening
Bug report, extracts developer's liveness information at the text information for extracting the Bug report filtered out.
Further, raw data set pre-process specifically using following steps:
S11: carrying out screening reserved state to bug report is the bug repaired, deletes the inefficient bug in vain and repaired；
S12: extracting text information from the bug report filtered out, chooses text information to be segmented, be gone at stop words
Reason deletes the excessive and very few word of frequency of occurrence；
S13: developer's liveness information is extracted, product and the identical developer of module information of current bug report are collected
Active sequences；
S14: pretreated data set is divided into training set, verifying collection and test set.
S2: carrying out pre-training to CSDBT model on training set, specifically in the following way:
S21: input uses One-Hot encoding textual information and developer's information；
S22: feature extraction is extracted the high-level characteristic of text information using bidirectional circulating neural network, uses one-way circulation
Neural network extracts the high-level characteristic of liveness information, and two kinds of high-level characteristics are merged in a manner of being multiplied between element；
S23: the conversion of high-level characteristic to developer's probability is completed；
S24: in training set data, to minimize cross entropy as target, optimization neural network model is extremely restrained.Wherein protect
The parameter constant of neural network is demonstrate,proved, new bug report is concentrated in input verifying, can return to developer's column by the big minispread of probability
Table.
S3: optimal misclassification cost matrix is solved using adaptive differential evolution algorithm: the data for verifying collection are asked
Solution obtains the optimal value of misclassification cost matrix.The purpose of the step be obtain an optimal misclassification cost matrix because
Each individual represents the feasible solution of a cost matrix in population, so we need to evaluate all feasible solutions,
To select best feasible solution, that is, select the highest individual of fitness.
Wherein cost sensitive learning, which refers to, provides different weights for different classes of sample, and core is misclassification cost square
Battle array.The misclassification cost of different classes of sample just constitutes misclassification cost matrix.Assuming that the number of classification is M, the data set
Corresponding misclassification cost matrix is exactly the matrix of a M × M:
Wherein Ci,jIndicate that the sample for actually belonging to i class is accidentally divided into cost caused by j class.
Under normal conditions, traditional training algorithm assume the cost of all misclassifications be all it is equal, i.e.,
But in practical applications, different classes of mistake often corresponds to different misclassification costs.Such as in medical treatment
In, the cost ＂ that Healthy People mistaken diagnosis is patient is different by ＂ by the cost ＂ and ＂ that patient's mistaken diagnosis is Healthy People certainly；Cost is quick
Sense study is exactly to consider different misclassification costs, distributes different costs for different types of mistake, so that in classification, it is high
The cost summation of quantity and mistake classification that cost mistake generates is minimum.By cost sensitive learning and neural network phase in the present invention
In conjunction with traditional neural network model, after classification, the posterior probability that a sample x belongs to classification i can be expressed as:
P (y=j | x)=softmaxi
Neural network will use the posterior probability that obtains in this way to recommend facilitate developer's list later.
But we additionally add a step, introduce cost-sensitive mechanism herein, and it is theoretical based on Bayes risk, it is general to posteriority
Rate is punished:
Then recommend facilitate developer's list with the new probability after this punishment by misclassification cost matrix above again.But
It is due to lacking priori knowledge, we can not learn the occurrence of the corresponding misclassification cost matrix of the data set, so we
It needs through differential evolution algorithm come optimization one optimal misclassification cost matrix.The misclassification generation that the feasible solution is represented
Valence matrix is combined with the good neural network of pre-training in above manner, with correct rate score of the model on verifying collection
Fitness value as the feasible solution.It is the evaluation method to a feasible solution above, if there is 100 individuals in population,
We need by the independent execution of the above process 100 times, to obtain fitness value of each individual.
S4: the good CSDBT model of pre-training and optimal misclassification cost matrix obtained in the previous step are combined to obtain
New CSDBT model, recently enters test set and tests new CSDBT model.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Anyone skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (5)
1. a kind of software bug allocating method based on Recognition with Recurrent Neural Network and cost-sensitive: the following steps are included:
S1: the raw data set in acquisition history bug report warehouse is pre-processed；Wherein pretreatment includes screening Bug report
It accuses, extract text information, extraction developer's liveness information that the Bug filtered out is reported；
S2: pre-training is carried out to CSDBT model on training set；
S3: optimal misclassification cost matrix is solved using adaptive differential evolution algorithm: the data for verifying collection are solved, and are obtained
Take the optimal value of misclassification cost matrix；
S4: the good CSDBT model of pre-training and optimal misclassification cost matrix obtained in the previous step are combined to obtain new
DeepTriage model recently enters test set and tests new CSDBT model.
2. a kind of software bug allocating method based on Recognition with Recurrent Neural Network and cost-sensitive according to claim 1, special
Sign also resides in: in S1 specifically in the following way:
S11: carrying out screening reserved state to bug report is the bug repaired, deletes the inefficient bug in vain and repaired；
S12: extracting text information from the bug report filtered out, chooses text information to be segmented, stop words is gone to handle,
Delete the excessive and very few word of frequency of occurrence；
S13: extracting developer's liveness information, collects product and the identical developer's activity of module information of current bug report
Sequence；
S14: pretreated data set is divided into training set, verifying collection and test set.
3. a kind of software bug allocating method based on Recognition with Recurrent Neural Network and cost-sensitive according to claim 1, special
Sign also resides in: in S2 specifically in the following way:
S21: One-Hot encoding textual information and developer's information are used；
S22: extracting the high-level characteristic of text information using bidirectional circulating neural network, is extracted and is lived using one-way circulation neural network
The high-level characteristic of jerk information merges two kinds of high-level characteristics in a manner of being multiplied between element；
S23: the conversion of high-level characteristic to developer's probability is completed；
S24: in training set data, to minimize cross entropy as target, optimization neural network model is extremely restrained.
4. a kind of software bug allocating method based on Recognition with Recurrent Neural Network and cost-sensitive according to claim 1, special
Sign also resides in: in S3 specifically in the following way:
S31: initialization misclassification cost population wherein has individual, each individual represents a misclassification cost square in population
The feasible solution of battle array；
S32: mutation operator and crossover operator based on adaptive generation generate a new misclassification cost population；
S33: calculating the fitness value of all individuals in population, the misclassification cost matrix that each individual represents is brought into respectively pre-
Trained CSDBT model, by the new CSDBT model of acquisition, operation obtains correct rate score on verifying collection, the accuracy number
Value is the fitness value of a feasible solution；
S34: according to current individual with the fitness value of the corresponding individual of previous generation, the higher individual of fitness is therefrom selected, is abandoned
Show the individual of difference；
S35: the mutation operator and crossover operator of adaptive adjustment population；
S36: it returns to S32 step and continues iteration, until reaching the maximum number of iterations of setting, and select fitness in last generation
Highest individual is used as optimal solution.
5. a kind of software bug allocating method based on Recognition with Recurrent Neural Network and cost-sensitive according to claim 1, special
Sign also resides in: in S4 specifically in the following way:
S41: on the CSDBT model for completing to instruct in advance, in conjunction with the misclassification cost matrix that optimization obtains, input is by pretreated
Test set；
S42:CSDBT model will return to nominator's list that a length is K for each sample in test set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811528909.9A CN109615242A (en) | 2018-12-13 | 2018-12-13 | A Recurrent Neural Network-Based and Cost-Sensitive Software Bug Assignment Method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811528909.9A CN109615242A (en) | 2018-12-13 | 2018-12-13 | A Recurrent Neural Network-Based and Cost-Sensitive Software Bug Assignment Method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109615242A true CN109615242A (en) | 2019-04-12 |
Family
ID=66008134
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811528909.9A Pending CN109615242A (en) | 2018-12-13 | 2018-12-13 | A Recurrent Neural Network-Based and Cost-Sensitive Software Bug Assignment Method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109615242A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110471854A (en) * | 2019-08-20 | 2019-11-19 | 大连海事大学 | A kind of defect report assigning method based on high dimensional data mixing reduction |
| CN111309907A (en) * | 2020-02-10 | 2020-06-19 | 大连海事大学 | Real-time Bug assignment method based on deep reinforcement learning |
| US11714743B2 (en) | 2021-05-24 | 2023-08-01 | Red Hat, Inc. | Automated classification of defective code from bug tracking tool data |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104009824A (en) * | 2014-06-01 | 2014-08-27 | 张喆 | Pilot assisted data fusion method based on differential evolution in base station coordination uplink system |
| CN107480141A (en) * | 2017-08-29 | 2017-12-15 | 南京大学 | It is a kind of that allocating method is aided in based on the software defect of text and developer's liveness |
-
2018
- 2018-12-13 CN CN201811528909.9A patent/CN109615242A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104009824A (en) * | 2014-06-01 | 2014-08-27 | 张喆 | Pilot assisted data fusion method based on differential evolution in base station coordination uplink system |
| CN107480141A (en) * | 2017-08-29 | 2017-12-15 | 南京大学 | It is a kind of that allocating method is aided in based on the software defect of text and developer's liveness |
Non-Patent Citations (1)
| Title |
|---|
| 蒋良孝，李超群编: "《贝叶斯网络分类器 算法与应用》", 31 December 2015 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110471854A (en) * | 2019-08-20 | 2019-11-19 | 大连海事大学 | A kind of defect report assigning method based on high dimensional data mixing reduction |
| CN110471854B (en) * | 2019-08-20 | 2023-02-03 | 大连海事大学 | A Defect Report Assignment Method Based on Hybrid Reduction of High-Dimensional Data |
| CN111309907A (en) * | 2020-02-10 | 2020-06-19 | 大连海事大学 | Real-time Bug assignment method based on deep reinforcement learning |
| US11714743B2 (en) | 2021-05-24 | 2023-08-01 | Red Hat, Inc. | Automated classification of defective code from bug tracking tool data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11500818B2 (en) | Method and system for large scale data curation | |
| US7685082B1 (en) | System and method for identifying, prioritizing and encapsulating errors in accounting data | |
| CN110111885B (en) | Attribute prediction method, attribute prediction device, computer equipment and computer readable storage medium | |
| CN110660478A (en) | Cancer image prediction and discrimination method and system based on transfer learning | |
| CN109615242A (en) | A Recurrent Neural Network-Based and Cost-Sensitive Software Bug Assignment Method | |
| US11763945B2 (en) | System and method for labeling medical data to generate labeled training data | |
| CN116383364A (en) | A medical question answering method and system based on doctor feedback and reinforcement learning | |
| CN106250311A (en) | Repeated defects based on LDA model report detection method | |
| Podgorelec et al. | Knowledge discovery with classification rules in a cardiovascular dataset | |
| CN112750529A (en) | Intelligent medical inquiry device, equipment and medium | |
| CN113723312A (en) | Visual transform-based rice disease identification method | |
| CN108920446A (en) | A kind of processing method of Engineering document | |
| CN111242565A (en) | Resume optimization method and device based on intelligent personnel model | |
| CN116431813A (en) | Intelligent customer service problem classification method and device, electronic equipment and storage medium | |
| Wang et al. | Coad: Automatic diagnosis through symptom and disease collaborative generation | |
| Dissanayake et al. | Two-level boosting classifiers ensemble based on feature selection for heart disease prediction | |
| CN113313470B (en) | Employment type assessment method and system based on big data | |
| CN118585428A (en) | A software testing method and device based on private knowledge base | |
| CN108491487A (en) | A kind of clinical guidelines knowledge encoding method and system | |
| Shi et al. | A general neural framework for classification rule mining | |
| CN118352089A (en) | A method and device for constructing a fundamental cause of death prediction model | |
| Li et al. | Empowering multi-class medical data classification by Group-of-Single-Class-predictors and transfer optimization: Cases of structured dataset by machine learning and radiological images by deep learning | |
| Bhatt | An Analytical Review of Deep Learning Algorithms for Stress Prediction in Teaching Professionals | |
| Hashim et al. | Advanced medical diagnosis and prediction using deep learning | |
| Semenov et al. | Implementation of a clinical decision support system for interpretation of laboratory tests for patients |
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: 20190412 |