CN107480721A - A kind of ox only ill data analysing method and device - Google Patents
A kind of ox only ill data analysing method and device Download PDFInfo
- Publication number
- CN107480721A CN107480721A CN201710717043.5A CN201710717043A CN107480721A CN 107480721 A CN107480721 A CN 107480721A CN 201710717043 A CN201710717043 A CN 201710717043A CN 107480721 A CN107480721 A CN 107480721A
- Authority
- CN
- China
- Prior art keywords
- ill
- data
- classification
- cost
- state index
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01K—ANIMAL HUSBANDRY; CARE OF BIRDS, FISHES, INSECTS; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
- A01K29/00—Other apparatus for animal husbandry
- A01K29/005—Monitoring or measuring activity, e.g. detecting heat or mating
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/22—Social work
Abstract
The embodiment of the present invention provides a kind of ox only ill data analysing method and device.This method specifically comprises the following steps:The state index data of ox health status can be characterized by obtaining;Sample data is selected, and sample training is carried out to selected sample data on the cost matrix of illness classification and more points of Bayes's process classification devices based on cost-sensitive according to what is initialized in advance, obtains the ill disaggregated model of training sample;According to obtained state index data, cost matrix and training pattern, calculate the ill class probability of ox to be measured and export.Method provided in an embodiment of the present invention is proposed based on the theoretical more points of Bayes's Gaussian processes of cost-sensitive of Bayes risk to handle the calculating of the ill class probability of ox only, the multiple state index parameters of ox only can be calculated and export multiple ill class probabilities, and erroneous judgement work factor is added, to overcome existing discrimination method only to pursue extensive low False Rate, the shortcomings that single classification results can only be exported.
Description
Technical field
The present invention relates to medical field, in particular to a kind of ox only ill data analysing method and device, for supervising
The ox for surveying cultivation is only particularly the health status of milk cow.
Background technology
The combination of current big data, technology of Internet of things and biological wearable device and application for milk cattle cultivating informationization,
It is scientific to provide important foundation and support.Classification, prediction scheduling algorithm are the technological means that ox is only commonly used in ill differentiation, are had
Relatively broad application.
Existing milk cow health differentiates that scheme generally use fixes threshold values or the mode of parameter is calculated, adjusting parameter or
Person's threshold values is required for re -training model, can not be dynamically adjusted in running.Current existing algorithm is in calculating process
In, extensive misdiagnosis rate is only taken into account, it is not too big to be likely to result in some indicator deviation, but is that risk factor is higher in fact
Index causes larger loss not by timely classification, exposure.In addition, the threshold parameter or weight of existing method more by
Animal doctor and industry specialists are disposably set or manually adjusted item by item, and accuracy rate is relatively low.
The content of the invention
In view of this, it is an object of the invention to provide a kind of ox only ill data analysing method and device, with improvement
State problem.
Present pre-ferred embodiments provide a kind of ox only ill data analysing method, for monitoring the health status of ox only,
Methods described includes:The state index data of ox health status can be characterized by obtaining;Sample data is selected, and according to first in advance
The cost matrix classified on illness of beginningization and more points of Bayes's process classification devices based on cost-sensitive are to selected sample
Notebook data carries out sample training, obtains the ill disaggregated model of training sample;According to obtained state index data, described
The ill disaggregated model of cost matrix and the training sample, calculate the ill class probability of the ox to be measured and export.
Further, methods described also includes:According to the actually detected result of the health status of the ox to be measured only and obtain
The ill class probability arrived, the cost matrix is updated based on AdaCost algorithms.
Further, the state index data include at least one in physical signs, sign Index, activity index
Class, wherein:The physical signs includes shell temperature, heart rate, body weight, the same day ruminates number, the same day ruminates the time, in appetite
It is at least one;The sign Index includes at least one of texture, form;The activity index includes day walking step number.
Further, the state index data, the cost matrix and the training sample that the basis obtains
Ill disaggregated model, the step of calculating the ill class probability of the ox to be measured and export, including:Calculate the state index
Posterior probability P (the c in ill classification of dataJ=1,2 ..., m|xI=1,2 ..., n), wherein:cjRepresent one in the ill classification
The ill classification of kind, xi∈ X={ x1,x2,...,xn, X represents the state index data set of the ox to be measured got, xiTable
Show a kind of state index data therein;According to calculating formula:Calculating refers to state
Mark data xiIt is predicted as ill classification cjCondition cost R (cj|xi), wherein, C represents the cost matrix, and C (k, j) is represented will
Ill classification ckMistake is determined as ill classification cjCost;According to formula:R(cji|xi)=minjR(cj|xi), determine xiIt is corresponding
Minimal condition cost, and and then determine state index data xiThe corresponding ill classification of prediction;According to above-mentioned identified knot
The ill disaggregated model of fruit and the training sample calculates the ill class probability of ox to be measured only;What output was calculated is directed to institute
State the ill class probability of ox to be measured only.
Further, the covariance kernel function on the more points of Bayes process classification device is gaussian kernel function.
Another preferred embodiment of the present invention provides a kind of ox only ill data analysis set-up, for monitoring the healthy shape of ox only
State, described device include:Data acquisition module, the state index data of ox health status can be characterized for obtaining;Model
Training module, for selecting sample data, and according to the cost matrix classified on illness initialized in advance and based on cost
More points of sensitive Bayes's process classification devices carry out sample training to selected sample data, obtain the illness point of training sample
Class model;Probability evaluation entity, for according to the obtained state index data, the cost matrix and the training sample
Ill disaggregated model, calculate the ill class probability of the ox to be measured only and simultaneously export.
Further, described device also includes:Cost matrix update module, for the healthy shape according to the ox to be measured only
The actually detected result of state and the obtained ill class probability, the cost matrix is carried out more based on AdaCost algorithms
Newly.
Further, the state index data include at least one in physical signs, sign Index, activity index
Class, wherein:The physical signs includes shell temperature, heart rate, body weight, the same day ruminates number, the same day ruminates the time, in appetite
It is at least one;The sign Index includes at least one of texture, form;The activity index includes day walking step number.
Further, the probability evaluation entity is according to the obtained state index data, the cost matrix and institute
The ill disaggregated model of training sample is stated, the mode for calculating the ill class probability of the ox to be measured and exporting includes:Calculate
Posterior probability P (the c in ill classification of the state index dataJ=1,2 ..., m|xI=1,2 ..., n), wherein:cjRepresent the trouble
A kind of ill classification in disease classification, xi∈ X={ x1,x2,...,xn, the state of the ox to be measured that X represents to get only refers to
Mark data set, xiRepresent a kind of state index data therein;According to calculating formula:
Calculate state index data xiIt is predicted as ill classification cjCondition cost R (cj|xi), wherein, C represents the cost matrix,
C (k, j) is represented ill classification ckMistake is determined as ill classification cjCost;According to formula:R(cji|xi)=minjR(cj|
xi), determine xiCorresponding minimal condition cost, and and then determination state index data xiThe corresponding ill classification of prediction;According to upper
The ill disaggregated model of result and the training sample determined by stating calculates the ill class probability of ox to be measured only;Output calculates
The obtained ill class probability for the ox to be measured only.
Ox provided in an embodiment of the present invention only ill data analysing method and device, it is proposed that theoretical based on Bayes risk
More points of Bayes's Gaussian processes of cost-sensitive come handle ox only be particularly milk cow ill class probability calculating.This method can
To be calculated to the multiple state index parameters of ox only and export multiple ill class probabilities, and add erroneous judgement cost because
Son, to overcome existing discrimination method only to pursue extensive low False Rate, the shortcomings that single classification results can only be exported.
Especially, this ox only by every state index data during milk cattle cultivating and be good for by ill data analysing method
Carry out dynamically associating calculating between Kang Wenti, for the various combination of various states achievement data, multiple target is exported by calculating
Class probability, and cost penalty factor is added, add high False Rate, the cost of high-risk sign Index erroneous judgement.By certainly
Dynamic study mechanism, and by being labeled to erroneous judgement result, the cost punishment parameter of specific erroneous judgement combined index is dynamically adjusted,
Ox only the differentiation accuracy of the doubtful problem of health, promptness rate are improved, to find individual ox health problem in cows in time, with
And the probability of the doubtful a variety of ill classification of output provides foundation.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below by embodiment it is required use it is attached
Figure is briefly described, it will be appreciated that the following drawings illustrate only certain embodiments of the present invention, therefore be not construed as pair
The restriction of scope, for those of ordinary skill in the art, on the premise of not paying creative work, can also be according to this
A little accompanying drawings obtain other related accompanying drawings.
Fig. 1 is a kind of side for being used to perform the ox only computing device of ill data analysing method provided in an embodiment of the present invention
Frame schematic diagram;
Fig. 2 is a kind of flow chart of ox provided in an embodiment of the present invention only ill data analysing method;
Fig. 3 is a kind of high-level schematic functional block diagram of ox provided in an embodiment of the present invention only ill data analysis set-up.
Icon:100- computing devices;110- oxen only ill data analysis set-up;120- memories;130- processors;112-
Data acquisition module;114- model training modules;116- probability evaluation entities;118- cost matrix update modules.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
Part of the embodiment of the present invention, rather than whole embodiments.The present invention implementation being generally described and illustrated herein in the accompanying drawings
The component of example can be configured to arrange and design with a variety of.
Therefore, below the detailed description of the embodiments of the invention to providing in the accompanying drawings be not intended to limit it is claimed
The scope of the present invention, but be merely representative of the present invention selected embodiment.It is common based on the embodiment in the present invention, this area
The every other embodiment that technical staff is obtained under the premise of creative work is not made, belong to the model that the present invention protects
Enclose.
Referring to Fig. 1, it is a kind of block diagram of computing device 100 provided in an embodiment of the present invention.Computing device 100
For performing the ox only ill data analysing method of the more points of Bayes procedures based on cost-sensitive, with the health problem to ox only
It is monitored.Instrument or sensor that the computing device 100 can be used to monitor ox characteristic index data with other etc. is connected,
With the achievement data collected;Or the computing device 100 can also receive some fingers of manual measurement for being manually entered
Mark data.
As shown in figure 1, computing device 100 comprises at least ox only ill data analysis set-up 110, memory 120 and place
Manage device 130.Wherein, ox only including at least one can be stored in the form of software or firmware deposit by ill data analysis set-up 110
In reservoir 120 or the software function module that is solidificated in the operating system of computing device 100.Processor 130 is used to perform storage
The executable module stored in device 120, such as ox software function module or computer that only ill data analysis set-up 110 includes
Program.
Referring to Fig. 2, it is a kind of flow chart for being used to monitor the method for ox health status provided in an embodiment of the present invention.
It should be noted that the method that the present embodiment provides is not using Fig. 2 and particular order as described below as limitation.Below will be to Fig. 2
Shown each step is described in detail.
Step S101, monitor the characteristic index data of ox to be measured only.In the present embodiment, various instruments, sensing can be passed through
Device or other smart machines monitor the various features achievement data of ox to be measured only.
Step S103, extract the state index data that ox health status can be characterized in characteristic index data.
In the present embodiment, a series of achievement datas that can characterize ox health status include, but not limited to physiology and referred to
It is at least a kind of in mark, sign Index, activity index.Wherein, physical signs is anti-including shell temperature, heart rate, body weight, the same day
Hay number, the same day ruminate at least one of time, appetite;Sign Index include but is not limited in texture, form at least one
Kind;Activity index includes but is not limited to day walking step number.
In the present embodiment, the state index data set extracted is denoted as X={ x1,x2,...,xn}。
In addition, by taking milk cow as an example, according to the practical experience during milk cattle cultivating, it is assumed that the ill category division of milk cow is
Following several classes:
Normally (N), limb hoof class disease (D1), internal organs class disease (D2), breeding class problem (D3)
As an example, the relation between the sample data and ill classification of milk cow state index data can use following mapping
Represent (assuming that):
The milk cow state index data of form 1 and ill classification mapping table
Step S105, sample data is selected, and according to the cost matrix classified on illness initialized in advance and be based on
More points of Bayes's process classification devices of cost-sensitive carry out sample training to selected sample data, obtain the trouble of training sample
Sick disaggregated model.
In the present embodiment, it is necessary to initialize the erroneous judgement cost square of ill classification in advance before being trained to sample data
Battle array.
By taking above-mentioned milk cow illness classification as an example, it is contemplated that be determined as healthy cow during milk cow health condition discrimination
Ill milk cow is different from the cost that ill milk cow is determined as to normal milk cow, therefore can construct the milk cow illness classification of below figure
Cost matrix table.Specifically, the cost of more serious or obvious disease state is arranged to 1.5, higher than normally or not
Determine state is arranged to 1.0, that is, it is expected that analysis result avoids illness being determined as normally as far as possible, cost matrix can be such as following table
It is shown:
The milk cow illness classification of form 2 judges cost matrix by accident
It is appreciated that it is above-mentioned be only by taking milk cow as an example, to associating between its common ill classification and state index data,
And the erroneous judgement cost between the different ill classifications that can be used in the present embodiment is illustrative but and non-limiting
's.
After initializing cost matrix, classification is trained to sample.
In the present embodiment, in order to avoid input sample data set deflection the problem of, choose sample data when, Various types of data
Proportion should be close.Because the characteristic of ill ox only is less compared to the data of overall group, therefore can all adopt
With.If the value of each state index data of sample data differs greatly, it usually needs sample data is standardized,
To reduce error.
In the present embodiment, if sample data training set is:
Wherein, S ∈ Rn, ciFor ill class label, N is total sample number, and n is the number of sample characteristics variable.
The covariance function of used more points of Bayes process classification device elects following height as when being trained to sample
This kernel function, because more points of Bayes's Gaussian process graders have good Nonlinear Classification performance.
Model learning is carried out based on above-mentioned sample training collection, obtains the ill disaggregated model of training sample.
Step S107, according to the ill disaggregated model of state index data, cost matrix and training sample, calculate ox to be measured
Ill class probability only simultaneously exports.
Specifically calculation can be:First calculate the posterior probability P in ill classification of state index data
(cJ=1,2 ..., m|xI=1,2 ..., n), wherein:cjRepresent a kind of ill classification in the ill classification, xi∈ X={ x1,
x2,...,xn, X represents the state index data set of the ox to be measured got, xiRepresent a kind of state index data therein;
Then according to calculating formula:Calculate state index data xiIt is predicted as ill class
Other cjCondition cost R (cj|xi), wherein, C represents the cost matrix, and C (k, j) is represented ill classification ckMistake is determined as
Ill classification cjCost;Further according to formula:R(cji|xi)=minjR(cj|xi), determine xiCorresponding minimal condition cost, and
And then determine state index data xiThe corresponding ill classification c of predictionji;Then, according to above-mentioned identified result and training sample
This ill disaggregated model calculates the ill class probability of ox to be measured only;Finally, export be calculated for ox to be measured only
Ill class probability.
Wherein, erroneous judgement cost usually requires the expert in cultivation field and provided according to conventional distinguishing rule or experience.But
It is that in a practical situation, the absolute figure of the erroneous judgement cost of each ill classification is difficult to provide, and judges cost C (k, j) essence by accident
The cost difference brought is judged by accident just for the sake of reacting between each illness is classified, therefore relative cost can be used in the present embodiment
To describe.The basic principle of the relative cost used in the method for the present embodiment is that the cost of correct decision is 0, the generation of erroneous judgement
Valency is more than 0, and has enough relative cost grades between minimum cost erroneous judgement and the erroneous judgement of maximum cost.
Preferably, in the method that the present embodiment provides, the step of can also including being updated cost matrix.For example,
According to actual conditions, the actually detected result of comprehensive veterinary or industry specialists, if the ill class probability result predicted
There is larger difference with actually detected result, then combination that can be to such state index is marked again, and adjustment ox is only ill
Cost matrix between classification.
Using AdaCost algorithms by iterating, a kind of finger-like state mark combination of each round iteration divides for the renewal of cost
Class device, and the sample weights of the performance renewal cost according to current class device.More new strategy is the sample weights drop correctly classified
Low, the classification samples weight for being noted as mistake increases, and final model is a weighted linear combination of successive ignition model,
The more accurate grader of classification will obtain bigger weight.
For correct indicator combination classification samples β+(i)=- 0.5Ci+ 0.5, it is noted as the classification indicators combination of mistake
Sample β-(i)=0.5Ci+ 0.5, the iteration of each round is calculated using equation below:
Wherein, β+(i) be work factor monotonic decreasing function, β-(i) in contrast, so that high cost sample β+
(i) it is low, and β-(i) it is high.
Method provided in an embodiment of the present invention relies only on the artificial side such as animal doctor's experience during having broken traditional milk cattle cultivating
The detection mode that formula is differentiated to milk cow health problem.The data that this method collects by external equipment or manually are to milk
Ox health automatically calculate and classify, and effectively raises the accuracy differentiated to milk cow health problem.It is in addition, of the invention
Embodiment can be supported to classify to the sign state index of the milk cow of input by bayesian algorithm of more classifying, while defeated
Go out the probability of a variety of different health problems.In addition, cost penalty mechanism is also added into, to milk cow physiology, sign, activity index
In high-risk index add cost penalty mechanism, improve the differentiation accuracy of high-risk index, be health during milk cattle cultivating
Problem, which differentiates, provides strong foundation.
Referring to Fig. 3, it is that the functional module of ox provided in an embodiment of the present invention only ill data analysis set-up 110 a kind of is shown
It is intended to.Only ill data analysis set-up 110 includes data acquisition module 112, model training module 114, probability evaluation entity to ox
116th, cost matrix update module 118.
The data acquisition module 112, the state index data of ox health status can be characterized for obtaining.The data
Acquisition module 112 can be used for performing the step S103 shown in Fig. 2.
The model training module 114, for selecting sample data, and according to initializing in advance on illness classification
Cost matrix and more points of Bayes's process classification devices based on cost-sensitive carry out sample training to selected sample data, obtain
To the ill disaggregated model of training sample.The model training module 114 can be used for performing the step S105 shown in Fig. 2, tool
The operating method of body is referred to above-mentioned elaborating on step S105.
The probability evaluation entity 116, for according to obtain the state index data, the cost matrix and described
The ill disaggregated model of training sample, calculate the ill class probability of the ox to be measured and export.The probability evaluation entity can
For performing the step S107 shown in Fig. 2, specific operating method is referred to above-mentioned explaining in detail on step S107
State.
The cost matrix update module 118, the actually detected result for the health status according to the ox to be measured only
With the obtained ill class probability, the cost matrix is updated based on AdaCost algorithms.
In summary, the embodiment of the present invention innovation the every state index proposed in a kind of breeding process by ox with
The computational methods that health problem is dynamically associated, it is directed to the various combination of many indexes, by calculating output multiple target point
Class probability, and cost penalty factor is added, add high False Rate, the cost of high-risk sign Index erroneous judgement.Also, by right
Erroneous judgement result is labeled, and dynamically adjusts the cost punishment parameter of specific erroneous judgement combined index, and improving ox, only health doubtful is asked
Rate of correct diagnosis, the promptness rate of topic, to find that individual ox health problem provides foundation in cows in time.
The foregoing is only a specific embodiment of the invention, but protection scope of the present invention is not limited thereto, any
Those familiar with the art the invention discloses technical scope in, change or replacement can be readily occurred in, should all be contained
Cover within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
It should be noted that:Similar label and letter represents similar terms in following accompanying drawing, therefore, once in a certain Xiang Yi accompanying drawing
It is defined, then it further need not be defined and explained in subsequent accompanying drawing.
Claims (10)
1. a kind of ox only ill data analysing method, it is characterised in that methods described includes:
The state index data of ox health status can be characterized by obtaining;
Sample data is selected, and according to more points on the cost matrix of illness classification and based on cost-sensitive initialized in advance
Bayes procedure grader carries out sample training to selected sample data, obtains the ill disaggregated model of training sample;
According to the ill disaggregated model of the obtained state index data, the cost matrix and the training sample, calculate
The ill class probability of the ox to be measured simultaneously exports.
2. ox according to claim 1 only ill data analysing method, it is characterised in that methods described also includes:
According to the actually detected result of the health status of the ox to be measured only and the obtained ill class probability, it is based on
AdaCost algorithms are updated to the cost matrix.
3. ox according to claim 1 only ill data analysing method, it is characterised in that the state index data include
It is at least a kind of in physical signs, sign Index, activity index, wherein:
The physical signs include shell temperature, heart rate, body weight, the same day ruminates number, the same day ruminates the time, in appetite at least
It is a kind of;
The sign Index includes at least one of texture, form;
The activity index includes day walking step number.
4. ox according to claim 1 only ill data analysing method, it is characterised in that the shape that the basis obtains
The ill disaggregated model of state achievement data, the cost matrix and the training sample, calculate the ox to be measured illness only point
Class probability and the step of export, including:
Calculate the posterior probability P (c in ill classification of the state index dataJ=1,2 ..., m|xI=1,2 ..., n), wherein:cjTable
Show a kind of ill classification in the ill classification, xi∈ X={ x1,x2,...,xn, X represents the ox to be measured got only
State index data set, xiRepresent a kind of state index data therein;
According to calculating formula:Calculate state index data xiIt is predicted as ill classification
cjCondition cost R (cj|xi), wherein, C represents the cost matrix, and C (k, j) is represented ill classification ckMistake is determined as suffering from
Sick classification cjCost;
According to formula:R(cji|xi)=minjR(cj|xi), determine xiCorresponding minimal condition cost, and and then determination state index
Data xiThe corresponding ill classification of prediction;
It is general that the ill classification of ox to be measured only is calculated according to the ill disaggregated model of above-mentioned identified result and the training sample
Rate;
Export the ill class probability for the ox to be measured only being calculated.
5. ox according to claim 1 only ill data analysing method, it is characterised in that on more points of Bayes's mistakes
The covariance kernel function of journey grader is gaussian kernel function.
6. a kind of ox only ill data analysis set-up, it is characterised in that described device includes:
Data acquisition module, the state index data of ox health status can be characterized for obtaining;
Model training module, for selecting sample data, and according to initialize in advance on illness classification cost matrix and
More points of Bayes's process classification devices based on cost-sensitive carry out sample training to selected sample data, obtain training sample
Ill disaggregated model;
Probability evaluation entity, for according to the obtained state index data, the cost matrix and the training sample
Ill disaggregated model, calculate the ill class probability of the ox to be measured and export.
7. ox according to claim 6 only ill data analysis set-up, it is characterised in that described device also includes:
Cost matrix update module, for the actually detected result of the health status according to the ox to be measured only and described in obtaining
Ill class probability, the cost matrix is updated based on AdaCost algorithms.
8. ox according to claim 6 only ill data analysis set-up, it is characterised in that the state index data include
It is at least a kind of in physical signs, sign Index, activity index, wherein:
The physical signs include shell temperature, heart rate, body weight, the same day ruminates number, the same day ruminates the time, in appetite at least
It is a kind of;
The sign Index includes at least one of texture, form;
The activity index includes day walking step number.
9. ox according to claim 6 only ill data analysis set-up, it is characterised in that the probability evaluation entity according to
The ill disaggregated model of obtained the state index data, the cost matrix and the training sample, calculate described to be measured
Ox only ill class probability and export mode include:
Calculate the posterior probability P (c in ill classification of the state index dataJ=1,2 ..., m|xI=1,2 ..., n), wherein:cjTable
Show a kind of ill classification in the ill classification, xi∈ X={ x1,x2,...,xn, X represents the ox to be measured got only
State index data set, xiRepresent a kind of state index data therein;
According to calculating formula:Calculate state index data xiIt is predicted as ill classification
cjCondition cost R (cj|xi), wherein, C represents the cost matrix, and C (k, j) is represented ill classification ckMistake is determined as suffering from
Sick classification cjCost;
According to formula:R(cji|xi)=minjR(cj|xi), determine xiCorresponding minimal condition cost, and and then determination state index
Data xiThe corresponding ill classification of prediction;
It is general that the ill classification of ox to be measured only is calculated according to the ill disaggregated model of above-mentioned identified result and the training sample
Rate;
Export the ill class probability for the ox to be measured only being calculated.
10. ox according to claim 6 only ill data analysis set-up, it is characterised in that on the more points of Bayes
The covariance kernel function of process classification device is gaussian kernel function.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710717043.5A CN107480721A (en) | 2017-08-21 | 2017-08-21 | A kind of ox only ill data analysing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710717043.5A CN107480721A (en) | 2017-08-21 | 2017-08-21 | A kind of ox only ill data analysing method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107480721A true CN107480721A (en) | 2017-12-15 |
Family
ID=60601607
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710717043.5A Pending CN107480721A (en) | 2017-08-21 | 2017-08-21 | A kind of ox only ill data analysing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107480721A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108766559A (en) * | 2018-05-22 | 2018-11-06 | 合肥工业大学 | Clinical decision support method and system for intelligent disorder in screening |
CN109102896A (en) * | 2018-06-29 | 2018-12-28 | 东软集团股份有限公司 | A kind of method of generating classification model, data classification method and device |
CN109686444A (en) * | 2018-12-27 | 2019-04-26 | 上海联影智能医疗科技有限公司 | System and method for medical image classification |
CN112185561A (en) * | 2020-09-28 | 2021-01-05 | 平安医疗健康管理股份有限公司 | User portrait generation method and device and computer equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103035050A (en) * | 2012-12-19 | 2013-04-10 | 南京师范大学 | High-precision face recognition method for complex face recognition access control system |
CN105930856A (en) * | 2016-03-23 | 2016-09-07 | 深圳市颐通科技有限公司 | Classification method based on improved DBSCAN-SMOTE algorithm |
CN106960218A (en) * | 2017-02-27 | 2017-07-18 | 同济大学 | Breast cancer data computer categorizing system based on cost sensitive learning Bayes |
-
2017
- 2017-08-21 CN CN201710717043.5A patent/CN107480721A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103035050A (en) * | 2012-12-19 | 2013-04-10 | 南京师范大学 | High-precision face recognition method for complex face recognition access control system |
CN105930856A (en) * | 2016-03-23 | 2016-09-07 | 深圳市颐通科技有限公司 | Classification method based on improved DBSCAN-SMOTE algorithm |
CN106960218A (en) * | 2017-02-27 | 2017-07-18 | 同济大学 | Breast cancer data computer categorizing system based on cost sensitive learning Bayes |
Non-Patent Citations (5)
Title |
---|
PEDRO DOMINGOS 等: "MetaCost: A General Method for Making Classifiers Cost-Sensitive", 《KDD "99 PROCEEDINGS OF THE FIFTH ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING》 * |
WEI FAN 等: "AdaCost: misclassification cost-sensitive boosting", 《ACM PROC OF THE 16TH INT CONF ON MACHINE LEARNING》 * |
XIAOYONG CHAI 等: "Test-cost sensitive naive bayes classification", 《FOURTH IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM"04)》 * |
曹鹏: "不均衡数据分类方法的研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
蒋盛益 等: "基于代价敏感的朴素贝叶斯不平衡数据分类研究", 《计算机研究与发展》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108766559A (en) * | 2018-05-22 | 2018-11-06 | 合肥工业大学 | Clinical decision support method and system for intelligent disorder in screening |
CN108766559B (en) * | 2018-05-22 | 2020-12-11 | 合肥工业大学 | Clinical decision support method and system for intelligent disease screening |
CN109102896A (en) * | 2018-06-29 | 2018-12-28 | 东软集团股份有限公司 | A kind of method of generating classification model, data classification method and device |
CN109686444A (en) * | 2018-12-27 | 2019-04-26 | 上海联影智能医疗科技有限公司 | System and method for medical image classification |
US11145405B2 (en) | 2018-12-27 | 2021-10-12 | Shanghai United Imaging Intelligence Co., Ltd. | Methods and devices for grading a medical image |
US11742073B2 (en) | 2018-12-27 | 2023-08-29 | Shanghai United Imaging Intelligence Co., Ltd. | Methods and devices for grading a medical image |
CN112185561A (en) * | 2020-09-28 | 2021-01-05 | 平安医疗健康管理股份有限公司 | User portrait generation method and device and computer equipment |
CN112185561B (en) * | 2020-09-28 | 2022-06-14 | 深圳平安医疗健康科技服务有限公司 | User portrait generation method and device and computer equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Al-Shayea | Artificial neural networks in medical diagnosis | |
CN107480721A (en) | A kind of ox only ill data analysing method and device | |
JP2023016929A (en) | Automated verification of medical data | |
CN112614133B (en) | Three-dimensional pulmonary nodule detection model training method and device without anchor point frame | |
Yin et al. | Estrus detection in dairy cows from acceleration data using self-learning classification models. | |
CN110767279A (en) | Electronic health record missing data completion method and system based on LSTM | |
CN106407664A (en) | Domain self-adaptive method and device of breathing gas diagnosis system | |
Pal | Identification of paddy leaf diseases using a supervised neural network | |
CN112256754A (en) | Ultrasonic detection analysis system and method based on standard model | |
KR102639558B1 (en) | Growth analysis prediction apparatus using bone maturity distribution by interest area and method thereof | |
JP2020190935A (en) | Machine learning program, machine learning method, and machine learning apparatus | |
CN109784645A (en) | A kind of hospital department performance arrangement method and device | |
KR102033484B1 (en) | Method and apparatus for setting normal reference range in clinical inspection of pets using generative adversary network | |
KR102187344B1 (en) | Method and apparatus for diagnosis in clinical inspection of pets using decision tree | |
CN116564521A (en) | Chronic disease risk assessment model establishment method, medium and system | |
US11961204B2 (en) | State visualization device, state visualization method, and state visualization program | |
CN109978877B (en) | Method and device for classifying by using screening model and storage medium | |
Verma et al. | Applicability of machine learning algorithms for intelligent farming | |
CN114814102B (en) | Rice bacterial leaf blight early warning method and device based on activity track sequence | |
Sivanesan et al. | Automated personalized health analytics using iot and machine learning algorithms | |
Sudhir et al. | Plant Disease Severity Detection and Fertilizer Recommendation using Deep Learning Techniques | |
CN117054354B (en) | Portable seed maturity spectrum detection system and device | |
WO2022209299A1 (en) | Information processing system, biological sample processing device, and program | |
CN117408823A (en) | Health data design life insurance scheme system for real-time human body monitoring | |
Zaylaa et al. | Early Diagnosis of Osteoporosis: An Artificial Intelligence-Based Framework |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
CB03 | Change of inventor or designer information |
Inventor after: Zhang Shuhua Inventor after: Yang Anrong Inventor after: Liu Lizhou Inventor after: Yi Jiangnan Inventor after: Wang Guodong Inventor before: Wang Guodong |
|
CB03 | Change of inventor or designer information | ||
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 |
Application publication date: 20171215 |
|
RJ01 | Rejection of invention patent application after publication |