CN110176291A - A kind of health and fitness information recommended method based on deep learning - Google Patents

A kind of health and fitness information recommended method based on deep learning Download PDF

Info

Publication number
CN110176291A
CN110176291A CN201910322542.3A CN201910322542A CN110176291A CN 110176291 A CN110176291 A CN 110176291A CN 201910322542 A CN201910322542 A CN 201910322542A CN 110176291 A CN110176291 A CN 110176291A
Authority
CN
China
Prior art keywords
user
deep learning
information
exercise
recommendation
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
Application number
CN201910322542.3A
Other languages
Chinese (zh)
Inventor
周凡
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sun Yat Sen University
Original Assignee
周凡
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 周凡 filed Critical 周凡
Priority to CN201910322542.3A priority Critical patent/CN110176291A/en
Publication of CN110176291A publication Critical patent/CN110176291A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/30—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to physical therapies or activities, e.g. physiotherapy, acupressure or exercising
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/60—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to nutrition control, e.g. diets

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Epidemiology (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)
  • Medical Informatics (AREA)
  • Theoretical Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Physical Education & Sports Medicine (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Nutrition Science (AREA)
  • Medical Treatment And Welfare Office Work (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)

Abstract

The health and fitness information recommended method based on deep learning that the invention discloses a kind of, this method comprises: the collection of information and arrangement, the pretreatment of information, enter information into obtain different foods to deep learning network model, from model recommend recommending index, choosing and recommend the highest several food of index and recommend the highest exercise program of index as final recommendation food and recommend amount of exercise recommendation plan for index and exercise program.The present invention combines deep learning with proposed algorithm, the feature of user and for further analysis is extracted from the wearable device or various relevant devices of user, it is excavated from the data of magnanimity and makes reliable, effective recommendation using effective information for user, according to the health data of user, reasonable food is made for user and amount of exercise is recommended.

Description

A kind of health and fitness information recommended method based on deep learning
Technical field
The present invention relates to deep learning and proposed algorithm technical fields, and in particular to a kind of health letter practised based on depth Cease recommended method.
Background technique
Deep learning originating from neural network research is a kind of high level more abstract by the formation of combination low-level feature Indicate attribute, algorithm to find data distribution formula character representation.The relatively common multilayer perceptron containing more hidden layers is exactly A kind of deep learning structure.As a kind of based on the method for carrying out representative learning to data, then deep learning in machine learning In, observation can be used various ways and be indicated.In general, can be easier using certain specific representation methods from example Middle learning tasks.As the frontier in a machine learning, deep learning is intended to establish a simulation human brain and carries out analytics The neural network of habit.It simulates the mechanism of human brain to explain data, and fruitful in certain field.
Proposed algorithm is then one of computer major by mathematics supposition user preference, provides the algorithm of suggestion.It calculates The work that method is substantially carried out is conceived to the feature in user model in the information of interest demand and destination item object model Information is matched, the primary dcreening operation calculated with relevant mathematical model, finds user's potentially interested recommendation items Mesh, and by project recommendation to user.As a kind of intelligent individual info service, proposed algorithm can pass through certain intelligence Generalization bounds realization is customized by targeted customized information.
A few days ago, as deep learning is increasingly becoming the hot spot of industry, both recommendation field is also just being thought deeply merge into each other can Can, although still the problems such as cold start-up, Sparse row exist in recommendation field, but with the addition of deep learning, algorithm The raising of energy great-leap-forward is not completely without possible.Time overhead is smaller, the higher proposed algorithm of accuracy just adding with deep learning Enter and is in the industry cycle fermented at leisure.
Existing technology has a kind of Chinese medicine medicinal material based on deep neural network of South China Science & Engineering University, production and research institute, Guangzhou to push away Recommend method, the specific steps are as follows:
1. the feature vector of computer acquisition human body tongue fur picture and corresponding Chinese medicine medicinal material digital code are as data set Input；
2. the feature vector of pair tongue fur picture carries out dimension-reduction treatment using Principal Component Analysis Algorithm；
3. the digital code of pair traditional Chinese medicine material carries out embedded characterization processing；
4. learning the relationship between tongue fur picture and Chinese medicine medicinal material；
5. calculating tongue fur figure using Chinese medicine medicinal material proposed algorithm according to the relationship between tongue fur picture and Chinese medicine requirement Degree of association score between piece and Chinese medicine medicinal material, and choose the high Chinese medicine medicinal material as recommendation of degree of association score.
The shortcomings that this method, is:
Traditional recommended method is limited to the collaborative filtering method of matrix decomposition, can only consider single score information.Separately Outside, as shallow Model, traditional algorithm causes to prejudge result and reality is managed without calligraphy learning to deeper characteristic information Thinking situation, there is biggish deviations.
Summary of the invention
The purpose of the present invention is overcoming the shortcomings of existing methods, proposes a kind of health and fitness information based on deep learning and push away Recommend method.Traditional recommended method is limited to the collaborative filtering method of matrix decomposition, can only consider single score information.Separately Outside, as shallow Model, traditional algorithm causes to prejudge result and reality is managed without calligraphy learning to deeper characteristic information Thinking situation, there is biggish deviations.The present invention proposes a kind of mixing interaction models based on deep learning, utilizes deep learning The model factor that more auxiliary informations are taken into consideration alleviates the excessively sparse problem of existing algorithm input data data, so It goes to learn more abstract profound character representation using the nonlinear network structure of multilayer interaction afterwards；By to user and its Profound information carries out the character representation result that multiple inner product interaction obtains more different levels；Finally it polymerize all knots Fruit is predicted.
To solve the above-mentioned problems, the health and fitness information recommended method based on deep learning that the invention proposes a kind of, it is described Method includes:
Weight, height, blood pressure, night vision feelings are obtained from smart machine (smartwatch, intelligent electronic-scale etc.) or user Condition, taste preference, lactose tolerance situation, the information of special allergic conditions.
The user information being collected into is passed through into one-hot coding, obtains sparse vector, then obtain corresponding feature vector.
Using the sparse vector of acquisition, feature vector as the input vector of neural network, it is input in network.
The recommendation index of various different foods and various amounts of exercise is calculated using the proposed algorithm in deep learning network (score), and according to preceding 5 food of size relation descending recommendation and first 3 amount of exercise plans for recommending index to user.
The health and fitness information recommended method based on deep learning that the invention proposes a kind of, this method by deep learning with push away It recommends algorithm to combine, the feature of user is extracted from the wearable device or various relevant devices of user and makees further point Analysis excavates from the data of magnanimity and makes reliable, effective recommendation using effective information for user, according to the strong of user Health data make reasonable food for user and amount of exercise are recommended.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, below will to embodiment or Attached drawing needed to be used in the description of the prior art is briefly described, it should be apparent that, the accompanying drawings in the following description is only Some embodiments of the present invention, for those of ordinary skill in the art, without creative efforts, also Other attached drawings can be obtained according to these attached drawings.
Fig. 1 is the health and fitness information recommended method flow diagram based on deep learning of the embodiment of the present invention；
Fig. 2 is the neural network configuration structure chart of the embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts it is all its His embodiment, shall fall within the protection scope of the present invention.
Fig. 1 is the health and fitness information recommended method flow diagram based on deep learning of the embodiment of the present invention, such as Fig. 1 institute Show, this method comprises:
S1 obtains weight, height, blood pressure, night from smart machine (smartwatch, intelligent electronic-scale etc.) or user Optionally, the information of taste preference, lactose tolerance situation, special allergic conditions.
The user information being collected into is passed through one-hot coding, obtains sparse vector, then obtain corresponding feature vector by S2.
S3, the sparse vector that S2 is obtained, feature vector are input in network as the input vector of neural network.
S4 calculates the recommendation of various different foods and various amounts of exercise using the proposed algorithm in deep learning network Index (score), and according to preceding 5 food of size relation descending recommendation and first 3 amount of exercise plans for recommending index to use Family.
Step S4 contains training stage and test phase in the practice of entire proposed algorithm.It is specific as follows:
The neural network configuration being related to such as Fig. 2.
In neural network, the number of plies of hidden layer N depends on the sum of food materials type and amount of exercise type.Define hidden layer It calculates as follows:
Z1=F1(W1x1)#(1)
Z2=F2(W2Z1+b2)#(2)
……
Zn=Fn(WnZn-1+bn)#(3)
N indicates the number of plies of neural network in hidden layer in formula (3).Define Zn,Fn,Wn,bnRespectively indicate the defeated of n-th layer Be worth out, the activation primitive of n-th layer, the weight matrix of n-th layer and the bias vector of n-th layer, bias vector regard concrete condition and It is fixed, enter no special case, defaults full 0.Activation primitive then selects tanh.
Interactive operation in each hidden layer calculates as follows:
Xu=Wuiu+Wuau′#(7)
Xv=Wuiv+Wvav′#(8)
pu1=F1(Wu1Xu)#(9)
qv1=F1(Wv1Xv)#(10)
……
pun=Fn(Wunpu(n-1)+bun)#(11)
qvn=Fn(Wunqv(n-1)+bvn)#(12)
Yn=pun⊙qvn#(13)
P is defined in formulaun、qvnThe output of n-th layer module respectively in hidden layer.Xu,XvRespectively indicate users personal data With food materials, amount of exercise data by the latent variable after embeding layer.WuiAnd WuaIndicate users personal data and food materials, movement Measure the weight matrix of data；WunIndicate the weight matrix of n-th layer in depth network training；⊙ indicates that corresponding element is multiplied；YnTable Show the interaction results that corresponding n-layer learns.
Entire method should include training stage and test phase.
Wherein the step of training stage includes:
[1] obtained original from smart machine (smartwatch, intelligent electronic-scale etc.) or user weight, height, Blood pressure, night vision situation, taste preference, lactose tolerance situation, special allergic conditions information input network, corresponding input is correct Food materials and amount of exercise information.
[2] inverted order rearrangement is converted to latent variable after carrying out one-hot coding to information, then according to tandem shown in Fig. 2.
[3] n is set, n should be the sum of food materials Yu amount of exercise type.
[4] vector after series connection is inputted into neural network, by available n different scores of hidden layer, corresponded to N kind food materials and amount of exercise.
[5] by obtained score compared with correct score, the mean square error L and gradient G of the two are calculated, L is defined such as Under:
WhereinIndicate that the score matrix of prediction, Y indicate true score matrix.
[6] each weight of the weight matrix of hidden layer subtracts gradient G.
[7] above-mentioned [4] to [6] process is repeated, until the value Jing Guo enough iteration or loss function is no longer obvious Until becoming smaller.
Test phase includes:
[1] obtained original from smart machine (smartwatch, intelligent electronic-scale etc.) or user weight, height, Blood pressure, night vision situation, taste preference, lactose tolerance situation, special allergic conditions information input network.
[2] obtained original from smart machine (smartwatch, intelligent electronic-scale etc.) or user weight, height, Blood pressure, night vision situation, taste preference, lactose tolerance situation, special allergic conditions information input network, corresponding input is correct Food materials and amount of exercise information.
[3] inverted order rearrangement is converted to latent variable after carrying out one-hot coding to information, then according to tandem shown in Fig. 2.
[4] n is set, n should be the sum of food materials Yu amount of exercise type.
[5] vector after series connection is inputted into neural network, by hidden layer available n different scores,
Correspond to n kind food materials and amount of exercise.
[6] all scores for being under the jurisdiction of food materials are distinguished into descending with the score for being under the jurisdiction of amount of exercise.Choose highest in food materials Point 5, in amount of exercise best result 3 (food materials and amount of exercise such as input are insufficient, then export corresponding export 1) conducts Recommendation.
The embodiment of the present invention propose a kind of health and fitness information recommended method based on deep learning, by deep learning with push away It recommends algorithm to combine, the feature of user is extracted from the wearable device or various relevant devices of user and makees further point Analysis excavates from the data of magnanimity and makes reliable, effective recommendation using effective information for user, according to the strong of user Health data make reasonable food for user and amount of exercise are recommended.
Those of ordinary skill in the art will appreciate that all or part of the steps in the various methods of above-described embodiment is can It is completed with instructing relevant hardware by program, which can be stored in a computer readable storage medium, storage Medium may include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), disk or CD etc..
In addition, be provided for the embodiments of the invention above a kind of health and fitness information recommended method based on deep learning into It has gone and has been discussed in detail, used herein a specific example illustrates the principle and implementation of the invention, the above implementation The explanation of example is merely used to help understand method and its core concept of the invention；Meanwhile for the general technology people of this field Member, according to the thought of the present invention, there will be changes in the specific implementation manner and application range, in conclusion this explanation Book content should not be construed as limiting the invention.

Claims (1)

1. a kind of health and fitness information recommended method based on deep learning, which is characterized in that the described method includes:
Weight, height, blood pressure, night vision situation, mouth are obtained from smart machine (smartwatch, intelligent electronic-scale etc.) or user Taste preference, lactose tolerance situation, the information of special allergic conditions.
The user information being collected into is passed through into one-hot coding, obtains sparse vector, then obtain corresponding feature vector.
Using the sparse vector of acquisition, feature vector as the input vector of neural network, it is input in network.
Recommendation index (of various different foods and various amounts of exercise is calculated using the proposed algorithm in deep learning network Point), and according to preceding 5 food of size relation descending recommendation and first 3 amount of exercise plans for recommending index to user.
CN201910322542.3A 2019-04-19 2019-04-19 A kind of health and fitness information recommended method based on deep learning Pending CN110176291A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910322542.3A CN110176291A (en) 2019-04-19 2019-04-19 A kind of health and fitness information recommended method based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910322542.3A CN110176291A (en) 2019-04-19 2019-04-19 A kind of health and fitness information recommended method based on deep learning

Publications (1)

Publication Number Publication Date
CN110176291A true CN110176291A (en) 2019-08-27

Family

ID=67689815

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910322542.3A Pending CN110176291A (en) 2019-04-19 2019-04-19 A kind of health and fitness information recommended method based on deep learning

Country Status (1)

Country Link
CN (1) CN110176291A (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106920147A (en) * 2017-02-28 2017-07-04 华中科技大学 A kind of commodity intelligent recommendation method that word-based vector data drives
CN107092797A (en) * 2017-04-26 2017-08-25 广东亿荣电子商务有限公司 A kind of medicine proposed algorithm based on deep learning
CN108182967A (en) * 2017-12-14 2018-06-19 华南理工大学 A kind of traditional Chinese medical science medicinal material based on deep neural network recommends method
CN108198603A (en) * 2017-12-12 2018-06-22 昆明亿尚科技有限公司 A kind of nutrition dietary based on personal physical examination information recommends method
CN108320187A (en) * 2018-02-02 2018-07-24 合肥工业大学 A kind of recommendation method based on depth social networks
WO2018201083A1 (en) * 2017-04-28 2018-11-01 University Of Southern California System and method for predicting survival time
CN109074867A (en) * 2016-04-25 2018-12-21 三星电子株式会社 Summarize the system and method for improving healthy result with successive learning for providing

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109074867A (en) * 2016-04-25 2018-12-21 三星电子株式会社 Summarize the system and method for improving healthy result with successive learning for providing
CN106920147A (en) * 2017-02-28 2017-07-04 华中科技大学 A kind of commodity intelligent recommendation method that word-based vector data drives
CN107092797A (en) * 2017-04-26 2017-08-25 广东亿荣电子商务有限公司 A kind of medicine proposed algorithm based on deep learning
WO2018201083A1 (en) * 2017-04-28 2018-11-01 University Of Southern California System and method for predicting survival time
CN108198603A (en) * 2017-12-12 2018-06-22 昆明亿尚科技有限公司 A kind of nutrition dietary based on personal physical examination information recommends method
CN108182967A (en) * 2017-12-14 2018-06-19 华南理工大学 A kind of traditional Chinese medical science medicinal material based on deep neural network recommends method
CN108320187A (en) * 2018-02-02 2018-07-24 合肥工业大学 A kind of recommendation method based on depth social networks

Similar Documents

Publication Publication Date Title
Opler Culture and social psychiatry
Klubertanz The philosophy of human nature
Wainwright Women healing/healing women: the genderisation of healing in early Christianity
Whitbourne et al. Adult development and aging: Biopsychosocial perspectives
Alexander Notes on the Synthesis of Form
Seymour-Smith Palgrave dictionary of Anthropology
CN106295186A (en) A kind of method and system of aided disease diagnosis based on intelligent inference
Malycha et al. The random-map technique: Enhancing mind-mapping with a conceptual combination technique to foster creative potential
CN108198604A (en) A kind of nutrition dietary based on personal characteristics recommends method
Pritchard et al. Knowledge and the politics of land
Zhan et al. Charting the developmental trajectories of attention and executive function in Chinese school-aged children
CN109920551A (en) Autism children social action performance characteristic analysis system based on machine learning
CN106844738A (en) The sorting technique of Junker relation between food materials based on neutral net
Hembry et al. Estimation of a nonlinear intervention phase trajectory for multiple-baseline design data
CN109933730A (en) A kind of excellent class of the master of education of middle and primary schools recommends resource application method and system
Barbosa Escobar et al. Crossmodal associations between visual textures and temperature concepts
Tyler Curriculum organization
Ryan et al. Existing semantic knowledge provides a schematic scaffold for inference in early cognitive decline, but not in amnestic MCI
Mehmood et al. Ramadan fasting increases judicial leniency in judges from Pakistan and India
CN110176291A (en) A kind of health and fitness information recommended method based on deep learning
Kim et al. An expert system for interpretation of structured mandala
CN112951374B (en) Data analysis method, system, computer device and storage medium
CN108764459A (en) Target identification network design method based on semantical definition
Rucinski Modelling learning to count in humanoid robots
de Paula Santil et al. Mental categorisation and classification process of information in maps

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
TA01 Transfer of patent application right
TA01 Transfer of patent application right

Effective date of registration: 20200807

Address after: 510006 No. 135 West Xingang Road, Guangzhou, Guangdong, Haizhuqu District

Applicant after: SUN YAT-SEN University

Address before: 510000 No. 29 Millennium Second Street, Haizhu District, Guangzhou City, Guangdong Province

Applicant before: Zhou Fan

RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20190827