AU2020286320B2 - Multi-granularity spark super trust fuzzy method applied to large-scale brain medical record segmentation - Google Patents
Multi-granularity spark super trust fuzzy method applied to large-scale brain medical record segmentation Download PDFInfo
- Publication number
- AU2020286320B2 AU2020286320B2 AU2020286320A AU2020286320A AU2020286320B2 AU 2020286320 B2 AU2020286320 B2 AU 2020286320B2 AU 2020286320 A AU2020286320 A AU 2020286320A AU 2020286320 A AU2020286320 A AU 2020286320A AU 2020286320 B2 AU2020286320 B2 AU 2020286320B2
- Authority
- AU
- Australia
- Prior art keywords
- granularity
- super
- population
- elites
- populations
- 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.)
- Active
Links
Classifications
-
- 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
- G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
- G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Primary Health Care (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Public Health (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Epidemiology (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
M J 4M 9W*#TrI Vf EP $il
(19)
P d PIT, R. ~(10) M WA/ NV4:Y
(43) d VTWO 2021/082444 A1
2021 4 5 ) 6 (06.05.2021) WIPO T PWC0T
(5 1) og-- J J(DING, Jialu); +l1 TA NMM 1 P
G16H 10/60 (2018.01) 9 , Jiangsu 226019 (CN)o .E It' (WANG,
(21) IgTh$-$ : PCT/CN2020/094104 Jiehua); + l I N d 1il ) N M 9
, Jiangsu 226019 (CN) o tf]l(HU,Bin); +[
(22) Ml8 i H: 2020 6 )] 3 H (03.06.2020) fl T @ M LM ) I I E N YG 9 l, Jiangsu 226019
(25) $ iFi : (CN)o 0tt* (CHEN, Senbo); + flV$l4
_,_J)Ia ns 22N 60, Jiangsu 226019 (CN)o 7TA (26)Qflii f : (WAN, Jie); + Il9dt) X N M 9
(30)VLR$R : Jiangsu226019 (CN)o kXITi(ZiAO,Lii); +[
201911030948.0 2019 4 10)28H (28.10.2019) CN fl inMdI u 2 2 609 , Jiangsu 226019
(71) $i :l5Mt4 (NANTONG UNIVERSITY) [CN/ (CN) o IJ (SUN,Ying); + Il W M T
CN]; + l T I N M d i )Il X IA N M )IZ X A N M 9 9, Jiangsu 226019 (CN)o 3 t
99, Jiangsu 226019 (CN)o (FENG, Zhihao); + [I M iliM ) IIEX N
9t, Jiangsu 226019 (CN)o kt(LI,Ming); +P
(72) &R ,: T -|V-7(DING, Weiping); + PId V$T IIMdTlTffIMid )l9 t N M9 Jiangsu 226019
i11)IlZAN M99,Jiangsu226019 (CN)o T (CN)o fiA (REN,Longjie); +[I3lW% Md
(54) Title: MULTI-GRANULATION SPARK-BASED SUPER-TRUST FUZZY METHOD FOR LARGE-SCALE BRAIN MEDI
CAL RECORD SEGMENTATION
-=(54)&TP t rrZr Fti n fl !-+tiSpark ~fl JMftM)t4
AA BB CC DD AA Divide a large-scale brain medical record data attribute set into multi-granulation evolutionary
SBB Design a multi-granulation Spark-based super-trust model to construct trust relations between
Ittt~9 - - l1F7& U
i~~uq~
vFl jJ~Tsuper elitists
CC Adjustiamulti-granulation Spark-based super-trust center threshold
H DID Execute supe elitist balance dynamicadustment strategytoenhance the optimal balanced
~k~] cnsistencofmulti-granulationsubopulaions
t~i~san. t~ F Store abrain medical recorddiagnosis caracteristic set on aSpark cloud platform
Tonassess the brainmeica record attriutecharacteristic segmentation precision or not?
EE FF GG GG Obtain the optimal large-scale
brain medical record characteristic set
HH Yes
11 No
1I)
(57) Abstract: A multi-granulation Spark-based super-trust fuzzy method for large-scale brain medical record segmentation, compris
ing: first, segmenting a large-scale brain medical record data attribute set into different multi-granulation evolutionary subpopulations
(Granu-populationi) on a Spark cloud platform; designing a multi-granulation Spark-based super-trust model to construct trust between
different super elitists in multi-granulation populations; adjusting a multi-granulation center threshold, and dynamically updating the
super elitists using a multi-granulation subpopulation balance adjustment strategy, performing global search segmentation and local
refinement segmentation on large-scale brain medical records, wherein super elitists can collaboratively extract knowledge reduction
subsets in respective regions; and finally, obtaining the optimal large-scale brain medical record segmentation characteristic set and
storing same on the Spark cloud platform. By means of the present method, stable segmentation can be implemented on large-scale
brain medical record knowledge reduction sets to provide important diagnostic basis for intelligent diagnosis and auxiliary treatment
of brain diseases.
(57) MR: -M f Tt!t)TJ v Nf 3#Pi !; +AMSpark{MIftrt29t)J , ticASpark 'T"li t tlft m
T MM) fI A$ M NT[ l] t i_;$+ $m 3f Tf fGranu-populationi4; itiT t-;+AM Spark@ ftry
r¾p REMR4 4 iQ , [lT V,; jRF t R 4) tl t r r F 1 ft SI-IE5-,' i[W4 T - i iSpark0 T +'&4 LL
wl t*tim~ vfu ni ttq#'v~L~94~ P
W O 2021/082444 A 1|11|||||||||||||||||||||||I||||||||||||||||||||||||||||||||i l i
TiT ) I I Z ' fW 9, Jiangsu 226019 (CN)
o
T
Ui)F(DING,Shuairong); t fl35fMl)II
E1 MO9, Jiangsu 226019 (CN)o
(74){IXi Ct ti] flR'
1] (NANJING JINGWEI PATENT & TRADEMARK
AGENCYCO.,LTD); +lT t1Z+
LiW17912tB), Jiangsu 210005 (CN)o
(8 1) 11- -M( M hr A, R V- PTJ) f-n(
Ef)f) : AE, AG, AL, AM, AO, AT, AU, AZ, BA, BB, BG,
BH, BN, BR, BW, BY, BZ, CA, CH, CL, CN, CO, CR, CU,
CZ, DE, DJ, DK, DM, DO, DZ, EC, EE, EG, ES, FI, GB,
GD, GE, GH, GM, GT, HN, HR, HU, ID, IL, IN, IR, IS,
JO, JP, KE, KG, KH, KN, KP, KR, KW, KZ, LA, LC, LK,
LR, LS, LU, LY, MA, MD, ME, MG, MK, MN, MW, MX,
MY, MZ, NA, NG, NI, NO, NZ, OM, PA, PE, PG, PH, PL,
PT, QA, RO, RS, RU, RW, SA, SC, SD, SE, SG, SK, SL,
ST, SV, SY, TH, TJ, TM, TN, TR, TT, TZ, UA, UG, US,
UZ, VC, VN, WS, ZA, ZM, ZWc
fi)): ARIPO (BW, GH, GM, KE, LR, LS, MW, MZ,
NA, RW, SD, SL, ST, SZ, TZ, UG, ZM, ZW), kXl (AM,
AZ, BY, KG, KZ, RU, TJ, TM), VIhlj (AL, AT, BE, BG,
CH, CY, CZ, DE, DK, EE, ES, FI, FR, GB, GR, HR, HU,
IE, IS, IT, LT, LU, LV, MC, MK, MT, NL, NO, PL, PT,
RO, RS, SE, SI, SK, SM, TR), OAPI (BF, BJ, CF, CG, CI,
CM, GA, GN, GQ, GW, KM, ML, MR, NE, SN, TD, TG).
- ( tt [ t yK - (*jKM21* (3))
Description
[0001] The present invention relates to the field of medical information, and specifically, to a multi-granularity Spark super trust fuzzy method applied to large-scale brain medical record segmentation.
[0002] For big data engineering of medical and health services, not only an electronic health record database and an electronic medical record database need to be constructed, but also a big data application system of medical and health administration and services covering public health, medical services, medical insurance, medicine supply, family planning, and general management services needs to be built. Under existing medical resource conditions, to achieve the objective of the big data engineering of medical and health services, a plurality of information technologies such as big data, cloud computing, and mobile Internet need to be fully utilized, to promote effective intercommunication between the electronic medical record database and the electrical health record database, and realize positive interaction to implement the big data engineering of medical and health services.
[0003] With the advent of cloud computing and big data era, large-scale electronic medical record intelligent processing is unusually complex in the entire medical big data generating and using processes, and medical data stored in an electronic medical record system has features such as large capacity, scattered sources, diverse formats, high access speed, and high application value. The key to forming a clinical decision support system is to effectively discover and extract important medical diagnosis rules and knowledge in the large-scale electronic medical records by using some artificial intelligence (Al) and data mining technologies. However, since the electronic medical record system is a special medical information system, the medical data stored in the electronic medical record system has complex features such as large capacity, variety, incompleteness, and timeliness, which bring relatively high difficulty in feature selection, collaborative services, knowledge discovery, and clinical decision support services. How to effectively deal with the large-scale complex electronic medical records is the key to designing future-oriented big data engineering of medical and health services and a clinical intelligent decision analysis and service system. A trend of future development is to perform reduction processing on the complex medical record knowledge by using some efficient models and methods in combination with the features of the large-scale electronic medical record system.
[0004] Brain attributes are automatically segmented from the large-scale brain medical record data by using Al and a big data processing method, which plays an important role in discovering potential medical rules, and prevention, control, and treatment of brain diseases. Segmentation issues of large-scale brain medical records widely exist in researches such as feature selection of brain medical record, rule mining, and clinical decision support system, which are core technologies of intelligent application of brain medical records under the background of medical big data. Therefore, it is urgent to consider to provide an effective method under the cloud computing environment to resolve the segmentation issues of the large-scale brain medical records, to further improve intelligent processing and service modes of massive brain medical records, which is a key problem to be urgently resolved in current researches of brain medical record intelligent-aided diagnosis and treatment and clinical decision support system under the background of medical big data, and is also a challenging research subject in the field of brain medical record. However, due to the high incompleteness and valuing fuzziness of the large-scale brain medical records, a feature of inauthenticity of attributes of the brain medical record data is more distinct and the uncertainty is more apparent, which greatly limit the application of the conventional attribute segmentation method. Therefore, in the environment of medical big data, putting forward an effective segmentation method for features of the large-scale brain medical records and obtaining an optimal consistent balance between global searching reduction and local refined knowledge collaborative reduction in the segmentation of the brain medical records are of great important meaning and value to decision support and analysis of the large-scale brain medical records.
Technical Problem
[0005] The present invention discloses a multi-granularity Spark super trust fuzzy method applied to large-scale brain medical record segmentation. The method includes the following steps: first segmenting attribute sets of large-scale brain medical record data to different multi-granularity evolutionary populations Granu-populationi on a Spark cloud platform; designing a multi-granularity-based Spark super trust model, and constructing trust degrees between different super elites in multi-granularity populations; adjusting a multi-granularity center threshold, dynamically updating the super elites by using a balanced adjustment strategy of multi-granularity population, and performing global searching segmentation and local refined segmentation on the large-scale brain medical records, so that the super elites can collaboratively extract knowledge reduction subsets in respective regions; and n F = UF finally calculating an optimal segmentation feature set =1 of the large-scale brain medical records and storing the optimal segmentation feature set to the Spark cloud platform. The present invention can stably segment a knowledge reduction set of the large-scale brain medical records, to provide an important diagnostic basis for intelligent diagnosis and adjuvant treatment of brain diseases.
Technical Solution
[0006] A further improvement of the present invention lies in that: specific steps of step B are as follows:
[0007] a. setting a quantity of the multi-granularity populations to n, where n>2, and initializing the multi-granularity populations to GPh where hE {1, ... , n};
[00081 b. initializing a center of a first granularity population to GSi, and then
initializing a center of a second granularity population to GSt used as a priority E, of a super elite;
[0009] c. for centers GS. of a third and more multi-granularity populations,
calculating a minimum distance between a current elite priority Ei and all centers of the current granularity populations, a calculation formula being as follows:
[0010] max min,, d(E', GS'))
[0011] allocating the minimum distance to a center GS of auhmulti-granularity population, and repeating this process until n multi-granularity evolutionary populations are all initialized;
[0012] d. defining a trust degree of an ith super elite in the same granularity population as follows:
dev(i)= 1 7jJ -SP
[00131 , where
[0014] n is a total quantity of elites, SPi is the ith super elite, and Pij is a jth common elite in an ith multi-granularity population;
[0015] e. calculating a trust degree Ri of the ith super elite SP in a center GS of an hth multi-granularity population, an iterative calculation formula being as follows:
/dev(T) x
[0016] 1ji(1de ,)where
[0017]iE {2,...,N},andR
[0018] f. setting a current quantity of times of cycles of a similarity between the centers GSh of the multi-granularity populations to t, where tE{2, ... , n-1}, and a
trust degree of each center GS; of the multi-granularity population is calculated through a previous cycle of a (t-1)th iteration, a size of the attribute sets of the large-scale brain medical records being dynamically iteratively updated by using a trust degree relationship between populations in different granular spaces;
[0019] g. calculating a trust difference Diffij between different super elites SPi and SPj in the multi-granularity populations, a calculation formula being as follows:
E Re -R, Dfff =m "j
[0020] , where
[0021] Reij is a trust degree of the ith super elite to a jth super elite, Rmj is a partial trust degree of an mth common elite randomly selected in a population recommended to the jth super elite, 1(j) is a setof all elites in a jth multi-granularity population GPj, and I(j)| is a cardinality of the set;
[0022] h. a population trust degree between centers of the hthmulti-granularity
population and the uth multi-granularity population being sim(GS ,GSI , a
calculation formula being as follows:
[0023] Sim(GShGS m ,where
[0024] m is a quantity of times of iterations, and A' (A) E [0,1 is a change range of a tth iteration of the two multi-granularity populations, and a calculation formula is
[0025] huR = sR -N
[0026] i. for the hth multi-granularity population GSh , if ]GS
. sim(GSh,GS'-) 6, 8 , being a similarity threshold, and a range being se[0, 1], the multi-granularity population meets the trust degree relationship between populations in different granular spaces; and
[0027] g. constructing a trust degree relationship formula between different super elites in the multi-granularity populations, the formula being defined as:
[02]Tri = A x Diff + (I- A)A' (R,), where 1002 81] ~her
[0029] 1 is a confidence factor of a direct trust degree between the super elites, a value of ) is related to a quantity of interactions between the super elites, and a greater quantity of interactions indicates a larger value of ), where 0 2<1; setting that k=h/HLmt, where h is a quantity of interactions between a super elite i and a super elite j, and HLmt is a specified threshold of the interaction quantity, and a size of the attribute sets of the large-scale brain medical records being dynamically iteratively updated by using a trust degree relationship between populations in different granular spaces.
[0030] A further improvement of the present invention lies in that: specific steps of step C are as follows:
[0031] a. initializing multi-granularity centers as by using a conventional clustering method k-means;
[00321 b. setting both a set of multi-granularity populations and the center to null sets, where V=<D and C=<D, and a quantity of times of iterations t is 1; and calculating a distance between each multi-granularity population and the multi-granularity center, dividing the attribute sets of the large-scale brain medical records to corresponding multi-granularity centers according to a minimum distance principle, to form k
{VO,} and record a quantity {NO,}i=1 of super elites in each center, and setting an
initial adjustment mark ioi=
[00331 c. recalculating each of the multi-granularity centers and an initial displacement d (ci, co) moved by the each granularity center, where |Vil represents a quantity of populations in the multi-granularity population Vi;
[0034] d. a distance between a granularity center ci of a granularity population after a first iteration and an initial granularity center co being d (ci, co), and a distance between a new granularity center c' after an ith iteration and an initial granularity d(c,c') center c being d(c, c'), where if d(cl,co) , c being a similarity threshold, and a range being cE[0, 1], the granularity center represented by c' no longer participates in a next cycle of iterative adjustment, otherwise the iterative adjustment is continued;
[0035] e. calculating distances between each super elite in a multi-granularity population of which a mark is ftj=1 and the centers of the multi-granularity populations participating in the adjustment, and dividing the brain medical record attributes to corresponding multi-granularity populations according to the minimum distance principle, to form k new multi-granularity populations {Vtj}, and record the quantity of super elites {Ntj} in each of the multi-granularity populations, to calculate the quantity ANtj of super elites configured to segment the large-scale brain medical record attributes after the adjustment;
[00361 f. recalculating the multi-granularity center = 'i participating in the adjustment and a displacement d (ctj, ctj) moved by the multi-granularity center; and
[00371 g. setting an adjustment threshold of a displacement of a granularity center to ,
and an adjustment threshold of a quantity of the multi-granularity populations to 0, and setting an adjustment mark in the multi-granularity centers Vi to 0 if the centers d(c,c ) C<0 ctj of the multi-granularity populations Vtj meet d(ce,c 2 1 ) and N namely, fij=, and adding Vtj and ctj to a final set of multi-granularity population centers, namely, V=V U {Vj} and C=C U {ctj}, where if a set including k multi-granularity centers is formed, |Vl=k, and the iteration is ended.
[0038] A further improvement of the present invention lies in that: specific steps of step E are as follows:
[0039] a. setting two adjacent super elite clusters as Ch and C.-', elite member
relation degrees of the two clusters being respectively 4 and
SI= E uM ,JrI
[00401b. if Sh > S , the super elites evolving into a combination of the elite cluster
Ch; otherwise, the super elites evolving into a combination of the elite cluster
C1, = P';
[0041] c. performing large-scale brain medical record segmentation with hybrid collaboration of competition and cooperation in the multi-granularity population, where assuming that Si is an ith super elite, the following operations are performed from i=1 to |Sil:
[0042] (1) inserting a representative Si,rep of the super elite Si into Pit;
[0043] (2) selecting the super elite Pit from the multi-granularity populations Granu-populationi ifnx>|Sil;
[0044] (3) combining solutions of all Sij and other multi-granularity populations Granu-populationi, sorting the values and calculating a minimum generated number of the Sij; and
[0045] (4) updating the representative of the super elite Si to obtain a non-dominated solution in the Pareto dominated region, deciding the multi-granularity population wining the selection, and updating Si=Sk;
[0046] d. calculating a fuzzy membership degree uCh (P) of the super elites through a similar membership manner, where a distance between a reference value Pi and a center Ch of the super elites is defined as d (Pi, Ch);
[00471 e. for each of the multi-granularity populations, a calculation balance CI of the
super elites is C nxn,and a consistent probability CR is D=(dnxn, where t El{l, 2, ... , s};
[00481 f. for any inconsistent balance degree C j)nxn, obtaining an optimal consistent balance degree of super elites of a tth multi-granularity population as
t nxn, where
[00491 ij , i, j, =1, 2, ... , n;
[00501 g. obtaining a global optimal consistent probability of all super elites as
) t nxn, where te {1, 2, ... , s}, and constructing a pair of the optimal consistent
balance degree and probability of the segmentation of the large-scale brain medical
record attributes as , >where tE {1, 2, ..., s}; and
[0051] h. the super elites segmenting, based on the pair 'of the optimal consistent balance degree and probability, feature sets of different attribute regions of the n
F =0 F medical records into F 1, F 2 , ... , F., and calculating an optimal feature set i=1
of the large-scale brain medical records.
Advantageous Effect
[0052] Compared with the prior art, the present invention has the following advantages.
[0053] 1) The present invention designs a multi-granularity-based Spark super trust model, constructs trust degrees between different super elites in a multi-granularity population, dynamically updates the super elites by using different balanced adjustment strategies of multi-granularity population, and performs global searching segmentation and local refined segmentation on the large-scale brain medical records, so that the super elites can collaboratively extract knowledge reduction subsets in respective regions, thereby greatly reducing execution time and improving segmentation precision of the large-scale brain medical records.
[0054] 2) The present invention constructs a dynamic collaborative operation mechanism of super elites in multi-granularity populations based on dynamic elite dominated regions on the Spark cloud platform, obtains an optimal consistent balance of the segmentation of the large-scale brain medical records, and reduces the complexity costs of the segmentation of features of the large-scale brain medical records. Further, the present invention improves fine granularity and robustness of parallel feature extraction of the large-scale brain medical records on the cloud computing Spark cloud platform, and lays a better foundation for development of smart services such as feature selection of brain medical records, rule mining, and clinical decision support.
[0055] FIG. 1 is an overview flowchart of a system;
[0056] FIG. 2 is a diagram of a dynamic execution process of a multi-granularity super trust Spark model; and
[0057] FIG. 3 to FIG. 5 are diagrams of a dynamic fuzzy collaborative operation process of super elites in multi-granularity populations.
[0058] To deepen the understanding of the present invention, the present invention is further described below in detail with reference to embodiments, and the embodiments are merely used for explaining the present invention and does not constitute a limitation to the protection scope of the present invention.
[0059] FIG. 1 to FIG. 5 show a specific implementation of a multi-granularity Spark super trust fuzzy method applied to large-scale brain medical record segmentation, and specific steps are as follows:
[0060] A. segmenting attribute sets of large-scale brain medical records to different multi-granularity evolutionary populations Granu-populationi, where i = 1, 2, . . n, to decompose a brain medical record attribute segmentation task into a plurality of parallel operation tasks, and then calculating equivalence classes of different candidate attribute sets of the brain medical records in the plurality of operation tasks obtained after the decomposition;
[00611 B. designing a multi-granularity-based super trust model, using an ih multi-granularity evolutionary population Granu-populationi to perform reduction and segmentation processing on an ith attribute set of the brain medical records, and constructing trust degrees between different super elites in multi-granularity populations to calculate a trust difference between the multi-granularity populations, a size of the attribute sets of the large-scale brain medical records being dynamically iteratively updated by using a trust degree relationship between populations in different granular spaces, where this step includes the following specific steps:
[0062] a. setting a quantity of the multi-granularity populations to n, where n>2, and initializing the multi-granularity populations to GPh where hE {1, ..., n};
[00631 b. initializing a center of a first granularity population to GS1', and then initializing a center of a second granularity population to GS2 used as a priority E of a super elite;
[0064] c. for centers GS of a third and more multi-granularity populations, calculating a minimum distance between a current elite priority Et and all centers of the current granularity populations, a calculation formula being as follows:
[00651 max (ninU, d(E|,GS.))
[0066] allocating the minimum distance to a center GS of a uh multi-granularity population, and repeating this process until n multi-granularity evolutionary populations are all initialized;
[0067] d. defining a trust degree of an ith super elite in the same granularity population as follows:
deV(T)
[00681 , where
[00691 n is a total quantity of elites, SPi is the ith super elite, and Pij is a jth common elite in an ith multi-granularity population;
[0070] e. calculating a trust degree Ri of the ith super elite SPi in a center GS of an hth multi-granularity population, an iterative calculation formula being as follows:
R, = - N dev(77) dvT) x Ri
[0071] j-i_(1/dev(7)) ,where
[0072] iE {2, ... , N}, and R, =r;
[0073] f. setting a current quantity of times of cycles of a similarity between the centers GSh of the multi-granularity populations to t, where te {2, ... , n-I}, and a
trust degree of each center GSh of the multi-granularity population is calculated through a previous cycle of a (t-1)th iteration, a size of the attribute sets of the large-scale brain medical records being dynamically iteratively updated by using a trust degree relationship between populations in different granular spaces;
[0074] g. calculating a trust difference Diffij between different super elites SPj and SPj in the multi-granularity populations, a calculation formula being as follows:
Z Re - Rm
[00751 I(j)j ,where
[0076] Reij is a trust degree of the ithsuper elite to a jth super elite, Rmj is a partial trust degree of an mth common elite randomly selected in a population recommended to the jth super elite, 1(j) is aset of all elites in a jthmulti-granularity population GPj, and I(j)| is a cardinality of the set;
[0077] h. a population trust degree between centers of the hthmulti-granularity
population and the uth multi-granularity population being sim(GS ,S') a calculation formula being as follows:
[00781 m , where
[0079] m is a quantity of times of iterations, and Ahu )E[0,1]isachangerangeof a tth iteration of the two multi-granularity populations, and a calculation formula is
[0080] A1u R= PsVi-P
[0081] i. for the hth multi-granularity population GSh, if AGS : sim(GSi, GSZ') 2, being a similarity threshold, and a range being cE [0, 1], the multi-granularity population meets the trust degree relationship between populations in different granular spaces; and
[0082] g. constructing a trust degree relationship formula between different super elites in the multi-granularity populations, the formula being defined as:
[0083] 8 1 , where
[0084] 1 is a confidence factor of a direct trust degree between the super elites, a value of X is related to a quantity of interactions between the super elites, and a greater quantity of interactions indicates a larger value of X, where 0 2 1; setting that X=h/HLmt, where h is a quantity of interactions between a super elite i and a super elite j, and HLmt is a specified threshold of the interaction quantity, and a size of the attribute sets of the large-scale brain medical records being dynamically iteratively updated by using a trust degree relationship between populations in different granular spaces.
[0085] C. setting an adjustment threshold of a multi-granularity Spark super trust center configured to segment the large-scale brain medical records to X, after an ih iteration is completed, performing a next iterative adjustment on a multi-granularity population Granu-populationi of which an adjustment amount of a granularity center is greater than the threshold ), setting an adjustment threshold of a displacement of the granularity center to , and an adjustment threshold of a quantity of the multi-granularity populations to 0, optimizing a center cij of a multi-granularity Vj, and adding the center to a final set of multi-granularity population centers, to form a set including k multi-granularity centers; and this step specifically includes the following steps:
[0086] a. initializing multi-granularity centers as{C}=1 by using a conventional clustering method k-means;
[0087] b. setting both a set of multi-granularity populations and the center to null sets, where V=C and C=, and a quantity of times of iterations t is 1; and calculating a distance between each multi-granularity population and the multi-granularity center, dividing the attribute sets of the large-scale brain medical records to corresponding multi-granularity centers according to a minimum distance principle, to form k
{i}=1, and record aquantity{No,}i=i of super elites in each center, and setting an initial adjustment mark foi =1 ;
[0088] c. recalculating each of the multi-granularity centers b and an initial displacement d (cu, coi) moved by the each granularity center, where |Vil represents a quantity of populations in the multi-granularity population Vi;
[0089] d. a distance between a granularity center ci of a granularity population after a first iteration and an initial granularity center co being d (ci, co), and a distance between a new granularity center c' after an ith iteration and an initial granularity d(c,c') center c being d(c, c'), where if d(c,,c) , being a similarity threshold, and a range being ce[0, 1], the granularity center represented by c' no longer participates in a next cycle of iterative adjustment, otherwise the iterative adjustment is continued;
[0090] e. calculating distances between each super elite in a multi-granularity population of which a mark is ftj=1 and the centers of the multi-granularity populations participating in the adjustment, and dividing the brain medical record attributes to corresponding multi-granularity populations according to the minimum distance principle, to form k new multi-granularity populations {Vj}, and record the quantity of super elites {Ntj} in each of the multi-granularity populations, to calculate the quantity ANtj of super elites configured to segment the large-scale brain medical record attributes after the adjustment;
c' x
[0091] f. recalculating the multi-granularity center =I # 1I participating in the adjustment and a displacement d (ctj, ctj) moved by the multi-granularity center; and
[0092] g. setting an adjustment threshold of a displacement of a granularity center to ,
and an adjustment threshold of a quantity of the multi-granularity populations to 0, and setting an adjustment mark in the multi-granularity centers Vi to 0 if the centers d(c__,c____ AN etj of the multi-granularity populations Vtj meet d(c,c 21 ) and Nt namely, fij=, and adding Vtj and etj to a final set of multi-granularity population centers, namely, V=V U {Vij} and C=C U {ctj}, where if a set including k multi-granularity centers is formed, |Vl=k, and the iteration is ended.
[0093] D. dynamically updating the super elites in the multi-granularity population by using a balanced adjustment strategy, to divide the super elites in the multi-granularity population to a right-angled isosceles triangle, to respectively calculate granularity values aNi of the super elites, where if two super elites have the same lower granularity aN3, similarity attribute values of the two super elites are converged into a balance pair (aN, av); and if the two super elites have the same higher granularity aN, the similarity attribute values of the two super elites are converged into a balance pair (N, aNi), the balanced adjustment strategy helping to increase the optimal consistent balance degree between the multi-granularity populations;
[0094] E. constructing a dynamic fuzzy collaborative segmentation strategy of the super elites in the multi-granularity population, performing global searching segmentation and local refined segmentation on the attributes of the large-scale brain medical records in a dynamic elite dominant region, to perform hybrid collaboration of competition and cooperation in the multi-granularity population, and constructing the optimal consistent balance degree and a probability level of the segmentation of the attributes of the large-scale brain medical records, so that the super elites collaboratively extract knowledge reduction subsets in respective corresponding Pareto dominant regions, and can stably segment different attribute regions of the
F=UF large-scale brain medical records, to calculate an optimal feature set i=1 of the large-scale brain medical records;
[0095] a. setting two adjacent super elite clusters as Ch and C.-', elite member
S' =( PCf relation degrees of the two clusters being respectively S, 1 =iIc and S = p (P1-)
[00961 b. if S>S, the super elites evolving into a combination of the elite cluster C; otherwise, the super elites evolving into a combination of the elite cluster
[0097] c. performing large-scale brain medical record segmentation with hybrid collaboration of competition and cooperation in the multi-granularity population, where assuming that Si is an ith super elite, the following operations are performed from i=1 to |Sil:
[0098] (1) inserting a representative Si,rep of the super elite Si into Pit;
[0099] (2) selecting the super elite Pit from the multi-granularity populations Granu-populationi if nx>|Sil;
[0100] (3) combining solutions of all Sij and other multi-granularity populations Granu-populationi, sorting the values and calculating a minimum generated number of the Sig; and
[0101] (4) updating the representative of the super elite Si to obtain a non-dominated solution in the Pareto dominated region, deciding the multi-granularity population wining the selection, and updating Si=Sk;
[0102] d. calculating a fuzzy membership degree uCh (Pi) of the super elites through a similar membership manner, where a distance between a reference value Pi and a center Ch of the super elites is defined as d (Pi, Ch);
[0103] e. for each of the multi-granularity populations, a calculation balance CI of the
super elites is C=(c nxn,and a consistent probability CR is D=(dnxn, where t E- {1, 2, ... , s}1;
[01041 f. for any inconsistent balance degreeC (c)nxn, obtaining an optimal consistent balance degree of super elites of a tth multi-granularity population as
C t)nxn, where
[01051 j , i,j,=1,2,...,n;
[0106] g. obtaining a global optimal consistent probability of all super elites as
= fxn, where tE {1, 2, ... , s}, and constructing a pair of the optimal consistent balance degree and probability of the segmentation of the large-scale brain medical
record attributes as ' D '>, where tE{1, 2, ... , s}; and
[0107] h. the super elites segmenting, based on the pair t of the optimal consistent balance degree and probability, feature sets of different attribute regions of the n F=UFl medical records into F1 , F 2 , ..., F,, and calculating an optimal feature set i=1 of the large-scale brain medical records.
[0108] F. comparing a relationship between the foregoing calculated segmentation precision RC of the large-scale brain medical records and a preset precision value 1, and if RC>q, outputting an optimal segmentation knowledge set of the large-scale brain medical records; otherwise, continuing to perform the foregoing steps C, D, and E, until the segmentation precision of the large-scale brain medical records meets the condition of RC>g; and n F=UF,
[0109] G. storing the optimal feature set i-1 of the segmentation of the large-scale brain medical records to the Spark cloud platform, to provide an important intelligent-aided diagnosis knowledge basis for clinical diagnosis and treatment of diseases related to the large-scale brain medical records.
[0110] The present invention designs a multi-granularity-based Spark super trust model, constructs trust degrees between different super elites in a multi-granularity population, dynamically updates the super elites by using different balanced adjustment strategies of multi-granularity population, and performs global searching segmentation and local refined segmentation on the large-scale brain medical records, so that the super elites can collaboratively extract knowledge reduction subsets in respective regions, thereby greatly reducing execution time and improving segmentation precision of the large-scale brain medical records.
[0111] The present invention constructs a dynamic collaborative operation mechanism of super elites in multi-granularity populations based on dynamic elite dominated regions on the Spark cloud platform, obtains an optimal consistent balance of the segmentation of the large-scale brain medical records, and reduces the complexity costs of the segmentation of features of the large-scale brain medical records. Further, the present invention improves fine granularity and robustness of parallel feature extraction of the large-scale brain medical records on the cloud computing Spark cloud platform, and lays a better foundation for development of smart services such as feature selection of brain medical records, rule mining, and clinical decision support.
[0112] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the present invention. Various modifications to these embodiments are obvious to a person skilled in the art, and the general principles defined in this specification may be implemented in other embodiments without departing from the spirit and scope of the present invention.
[0113] Therefore, the present invention is not limited to these embodiments illustrated in this specification, but needs to conform to the broadest scope consistent with the principles and novel features disclosed in this specification.
Claims (3)
1. A multi-granularity super trust fuzzy method applied to large-scale brain medical record
segmentation to obtain optimal feature sets for use in a clinical decision support system, comprising
the following specific steps:
A. segmenting a plurality of large-scale brain medical records to n different multi-granularity
evolutionary populations, Granu-populationi wherein i = 1, 2, ... n, and n > 2 and each Granu
populationi comprises a plurality of attribute sets and each attribute set has a size and each Granu
populationi is independently generated;
B. generating a multi-granularity-based super trust model configured to iteratively perform reduction
and segmentation processing on the plurality of attribute sets of the brain medical records in each
Granu-populationi, wherein the model is configured to calculate a trust degree between a super elite
and each common elite in a multi-granularity populations for each multi-granularity populations, and
to calculate a trust difference between a super elite in a first multi-granularity population and a super
elite in a second multi-granularity population and a trust degree relationship between super elites in
the two multi-granularity populations based on the calculated trust differences and trust degrees, and
the model where E E [0, 1], wherein the trust degree relationship is based on k which is a confidence
factor of a direct trust degree between a pair of super elites and is related to a quantity of interactions
between the super elites, wherein a common elite is an attribute set in a Granu-population and a super
elite is another attribute set in the same Granu-population wherein the super elite has a minimum
distance to a center of the Granu-population
C. generating a set V of multi-granularity populations and a set C of centers of the multi-granularity
populations and iteratively adjusting one or more centers of the multi-granularity populations using an adjustment threshold to optimize a center ctj of a jth multi-granularity populations in multi-granularity set Vtj to obtain a final set of k multi-granularity population centers, by a. using a k-means clustering method to determine initial centres{Co} 1 for each multi-granularity population; b. in a first iteration, t=1, setting a set, V, of multi-granularity populations to a null set and a set center C to a null set; and calculating a distance between each multi-granularity population and the multi-granularity center, dividing the attribute sets of the large-scale brain medical records to the nearest multi-granularity center according to a minimum distance formula to form k{Vt} 1, and record a quantity{No0 }_1 of super elites in each center, and setting an initial adjustment mark
(foilk=1 = {1}
c. recalculating each of the multi-granularity centers C1 = jZXEV xandaninitialdisplacement d
(ci, co) moved by the each granularity center, wherein|Vil represents a quantity of populations in the
multi-granularity population Vi;
d. calculating a distance between a granularity center ci of a multi-granularity population after a first
iteration and an initial granularity center co, d (ci, co), and calculating a distance between a new
granularity center c' after an i* iteration and an initial granularity center c, d(c, c'), wherein if a(c'cr> d(ci,c0 )
E, the granularity center represented by c'no longer participates in a next cycle of iterative adjustment
of the one or more centres of the multi-granularity populations, otherwise the iterative adjustment is
continued;
e. calculating distances between each super elite in a multi-granularity population of which a mark is
fj=1 and the centers of the multi-granularity populations participating in the adjustment, and dividing
the brain medical record attributes to corresponding multi-granularity populations according to a
minimum distance formula, to form k new multi-granularity populations {Vj}, and recording the quantity of super elites {N } in each of the multi-granularity populations, and calculating a quantity
ANj of super elites configured to segment the large-scale brain medical record attributes after the
adjustment;
1 f. recalculating the multi-granularity centerCtj = FVZXEV x participating in the adjustment and a
displacement d (cj, ej) moved by the multi-granularity center; and
g. setting an adjustment threshold of a displacement of a multi-granularity center to , and an
adjustment threshold of a quantity of the multi-granularity populations to 0, and setting an adjustment
mark in the multi-granularity centers Vj to 0 if the centers cj of the multi-granularity populations Vj
meet d(ct 1 c ) 1 < E and ANtj < 0, namely, fj=0, and adding Vj and cj to a final set of multi-granularity d(c 1 ,c 21 ) Ntj
population centers, namely, V = V U {Vt} and C = C U {ct 1}, wherein if a set including k multi
granularity centers is formed, IVl=k, and the iteration is ended.
D. dynamically updating the super elites in the multi-granularity population by using a balanced
adjustment strategy, in which the super elites in the multi-granularity population are projected into a
right-angled isosceles triangle based on granularity values aNi of the super elites wherein the
granularity values are the trust degree of the super elites, wherein if two super elites have the same
lower granularity aN 3 , similarity attribute values of the two super elites, that is the trust difference
Diffl between two super elites, are converged into a balance pair (aN 3 ,aN 3 ); and if the two super
elites have the same higher granularity aNi, the similarity attribute values of the two super elites are
converged into a balance pair (aN1 aN1 ) wherein the balanced adjustment strategy is configured to
increases an optimal consistent balance degree between the multi-granularity populations;
E. performing a dynamic fuzzy collaborative segmentation strategy of the super elites in the multi
granularity population to calculate an optimal feature set n
F = UF i=1
of the large-scale brain medical records by:
a. calculating a sum of the elite membership to two adjacent super elite clusters Ch and
C.-lusingS = z1-ipc, (P') and S.-1 = ipc-1 (P');
b. if S' > S'-1, then the super elites evolving into a combination of the cluster Ch; otherwise, the
super elites evolving into a combination of the cluster C = PCt;
c. performing hybrid collaboration of competition and cooperation in the multi-granularity
population such that different elites in the same granularity evolutionary population perform
competition of attribute reduction and different super elites in different granularity evolutionary
populations perform cooperation of attribute reduction, wherein assuming that Si is an ith super elite,
the following operations are performed from i=1 to | Si I:
(1) inserting a representative Si,p of the super elite Si into Pi;
(2) selecting the super elite Pi from the ith multi-granularity population Granu-populationi if n,>
Si;
(3) combining solutions of all SiJ and other multi-granularity populations Granu-populationi,
sorting the values and calculating a minimum generated number of the SiJ; and
(4) updating the representative of the super elite Si to obtain a non-dominated solution in the
Pareto dominated region, determining the winning multi-granularity population Sk, and
updating Si = Sk; d. calculating a fuzzy membership degree uCf (Pi) of each super elite to granupopulationi using a similarity measure, wherein a distance between a reference value Pi and a center Ch of the super elite is defined as d(Pi, Ch); e. for each of the multi-granularity populations, a calculation balance CI of the super elites is
Ct = (ct;),x , and a consistent probability CR is D = (dig),, wherein t E {1, 2, . . s};
f. for any inconsistent balance degree Ct = (ct)...obtaining an optimal consistent balance degree
of super elites of a t multi-granularity population as C = (Et;)""", wherein etj= and
ij, = 1, 2, ... , n;
g. obtaining a global optimal consistent probability of all super elites as Dt = (dt )"", wherein
t E {1, 2, . . s}, and constructing a pair of the optimal consistent balance degree and probability of the
segmentation of the large-scale brain medical record attributes astt < et, >, wherein t E
{1, 2, . . s}; and
h. segmenting the super elites into feature sets of different attribute regions of the medical records
into F 1, F 2 , . . , Fn, based on the pair it of the optimal consistent balance degree and probability, and
calculating the optimal feature set
F= Fi i=1
of the large-scale brain medical records;
F. estimating a segmentation precision RC based on the optimal feature set of step E and comparing
with a preset precision value I, and if RC > T then outputting the optimal segmentation knowledge set
of the large-scale brain medical records; otherwise, continuing to perform the foregoing steps C, D,
and E, until the segmentation precision meets the condition of RC ;> I; and
G. storing the optimal feature set
n
F = UF i=1
of the segmentation of the large-scale brain medical records to a cloud platform, wherein in use, a
clinical decision support system uses the optimal feature sets of the large-scale brain medical records
to analyse the large-scale brain medical records to provide clinical diagnosis and treatment of diseases
related to the large-scale brain medical records.
2. The multi-granularity super trust fuzzy method applied to large-scale brain medical record
segmentation according to claim 1, wherein specific steps of step B comprise:
a. initializing a centre GPf for each multi-granularity populations, wherein h E {1,..., n} comprising:
initializing a center of a first granularity population to GSf, and then initializing a center of a second
granularity population to GSt to an elite Et used as a super elite, and
for each remaining multi-granularity population, determining a center GSt of a u' multi-granularity
population by calculating the elite Ef with a minimum distance between the elite Ef and all centers of
the current granularity populations according to a minimum distance formula:
max (min d(E, GSf)), t u<h /
and the elite with the minimum distance is the super elite of the u' multi-granularity population, such
that after repeating this process each of the n multi-granularity evolutionary populations are all
initialized;
b. calculating a trust degree of an i"' super elite in the same granularity population is calculated using:
dev(Ti)= - 1 |Pi; SPI|, 2 j3 wherein q is a total quantity of elites, SPi is thei"' super elite, and Pij is a j' common elite in an id multi-granularity population; c. calculating a trust degree Ri of the i"' super elite SPi in a center GSt of an h"' multi-granularity population is calculated using an iterative calculation formula as follows: dev(Ti) Ey1 1/dev(T) wherein i E {2 . . N}, and R1 = V; d. calculating a trust difference Diffij between different super elites SPi and SPj in the multi granularity populations is calculated using :
Diff- -= f meI(j)Rej; - Rmj|
wherein Reij is a trust degree of the i"' super elite to a jt' super elite, R.j is a partial trust degree of an
m common elite randomly selected in a population recommended to the j' super elite, 1(j) is a set
of all elites in a j' tmulti-granularity population GPj, and I1(j) is a cardinality of the set;
e. calculating a population trust degree between centers of the ht" multi-granularity population and
the ut' multi-granularity population, sim(GSt, GSt), according to:
Em, A.(Ri) sim(GS, GSt) = 1- ,
m
wherein m is a quantity of times of iterations, and At,(Ri) E [0,1] is a change range of a t" iteration
of the two multi-granularity populations calculated according to
hu(Ri) = |pGS ~(I-(P') ; f. iteratively updating a size of the attribute sets in the h"' Granu-population if a center of the Granu population is similar to a center of Granu-population in a previous iteration by determining, for the h" multi-granularity population GSt, if 3GS- 1: sim(GS GS'-) f, E in which case the h"' multi granularity population meets the trust degree relationship with Granu-population in a previous iteration; and g. constructing a trust degree relationship formula between different super elites in the multi granularity populations according to:
Tri = A x Diffi + (1 - A)At (Ri),
wherein k=h/HLt, wherein h is a quantity of interactions between a super elite i and a super elite j, and
HLm.tis a specified threshold of the interaction quantity.
3. The multi-granularity super trust fuzzy method as claimed in claim 1 or 2, wherein the method is
performed using an Apache Spark cloud computing platform.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911030948.0A CN110867224B (en) | 2019-10-28 | 2019-10-28 | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation |
| CN201911030948.0 | 2019-10-28 | ||
| PCT/CN2020/094104 WO2021082444A1 (en) | 2019-10-28 | 2020-06-03 | Multi-granulation spark-based super-trust fuzzy method for large-scale brain medical record segmentation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| AU2020286320A1 AU2020286320A1 (en) | 2021-05-27 |
| AU2020286320B2 true AU2020286320B2 (en) | 2022-10-20 |
Family
ID=69653442
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| AU2020286320A Active AU2020286320B2 (en) | 2019-10-28 | 2020-06-03 | Multi-granularity spark super trust fuzzy method applied to large-scale brain medical record segmentation |
Country Status (3)
| Country | Link |
|---|---|
| CN (1) | CN110867224B (en) |
| AU (1) | AU2020286320B2 (en) |
| WO (1) | WO2021082444A1 (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110867224B (en) * | 2019-10-28 | 2022-02-08 | 南通大学 | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation |
| CN113012775B (en) * | 2021-03-30 | 2021-10-08 | 南通大学 | Incremental attribute reduction Spark method for classifying red spot electronic medical record pathological changes |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110867224A (en) * | 2019-10-28 | 2020-03-06 | 南通大学 | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation |
Family Cites Families (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN201788510U (en) * | 2010-07-13 | 2011-04-06 | 南通大学 | Dynamic EMR collaborative mining system with particle swarm and extension rough set/concept lattice theories integrated together |
| US8489526B2 (en) * | 2010-11-24 | 2013-07-16 | International Business Machines Corporation | Controlling quarantining and biasing in cataclysms for optimization simulations |
| EP2784748B1 (en) * | 2013-03-28 | 2017-11-01 | Expert Ymaging, SL | A computer implemented method for assessing vascular networks from medical images and uses thereof |
| CN103838972B (en) * | 2014-03-13 | 2016-08-24 | 南通大学 | A kind of quantum coordinating game model implementation method for MRI case history attribute reduction |
| CN104462853B (en) * | 2014-12-29 | 2017-05-24 | 南通大学 | Population elite distribution cloud collaboration equilibrium method used for feature extraction of electronic medical record |
| CN105069503A (en) * | 2015-07-30 | 2015-11-18 | 重庆邮电大学 | Cooperation degree-based hetero-population parallel particle swarm algorithm and implementation method of MapReduce model |
| CN105279388B (en) * | 2015-11-17 | 2017-11-17 | 南通大学 | The pregnant age newborn baby case history of multilayer cloud computing framework collaboration integrates reduction method |
| CN105719004A (en) * | 2016-01-18 | 2016-06-29 | 合肥工业大学 | Coevolution-based particle swarm optimization for solving multitask problems |
| CN106157370B (en) * | 2016-03-03 | 2019-04-02 | 重庆大学 | A kind of triangle gridding normalization method based on particle swarm algorithm |
| US20180108430A1 (en) * | 2016-09-30 | 2018-04-19 | Board Of Regents, The University Of Texas System | Method and system for population health management in a captivated healthcare system |
| CN109120017A (en) * | 2017-06-22 | 2019-01-01 | 南京理工大学 | A kind of Method for Reactive Power Optimization in Power based on improvement particle swarm algorithm |
| CN107257307B (en) * | 2017-06-29 | 2020-06-02 | 中国矿业大学 | Spark-based method for solving multi-terminal cooperative access network by parallelization genetic algorithm |
| CN108133260B (en) * | 2018-01-17 | 2021-05-07 | 浙江理工大学 | Real-time state monitoring-based multi-target particle swarm optimization workflow scheduling method |
| CN108446740B (en) * | 2018-03-28 | 2019-06-14 | 南通大学 | A kind of consistent Synergistic method of multilayer for brain image case history feature extraction |
| CN108986872B (en) * | 2018-06-21 | 2021-05-07 | 南通大学 | Multi-granularity attribute weight Spark method for big data electronic medical record reduction |
| CN109117864B (en) * | 2018-07-13 | 2020-02-28 | 华南理工大学 | Coronary heart disease risk prediction method, model and system based on heterogeneous feature fusion |
| CN109840551B (en) * | 2019-01-14 | 2022-03-15 | 湖北工业大学 | Method for optimizing random forest parameters for machine learning model training |
| CN109871995B (en) * | 2019-02-02 | 2021-03-26 | 浙江工业大学 | Quantum optimization parameter adjusting method for distributed deep learning under Spark framework |
-
2019
- 2019-10-28 CN CN201911030948.0A patent/CN110867224B/en active Active
-
2020
- 2020-06-03 WO PCT/CN2020/094104 patent/WO2021082444A1/en active Application Filing
- 2020-06-03 AU AU2020286320A patent/AU2020286320B2/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110867224A (en) * | 2019-10-28 | 2020-03-06 | 南通大学 | Multi-granularity Spark super-trust fuzzy method for large-scale brain pathology segmentation |
Non-Patent Citations (1)
| Title |
|---|
| DING WEIPING et al., 'Co-evolutionary cloud-based attribute ensemble multi-agent reduction algorithm', JOURNAL OF SOUTHEAST UNIVERSITY (ENGLISH EDITION), 15 December 2016, vol. 32, no. 4, p. 432-438 * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021082444A1 (en) | 2021-05-06 |
| CN110867224B (en) | 2022-02-08 |
| CN110867224A (en) | 2020-03-06 |
| AU2020286320A1 (en) | 2021-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Keuper et al. | Motion segmentation & multiple object tracking by correlation co-clustering | |
| US11481456B2 (en) | Model and pattern structure online unital learning: mapsoul | |
| Wang et al. | DeepSD: Supply-demand prediction for online car-hailing services using deep neural networks | |
| US7478104B2 (en) | Information processing apparatus, information processing method, and information processing program | |
| CN112308157B (en) | Decision tree-oriented transverse federated learning method | |
| CN110728317A (en) | Training method and system of decision tree model, storage medium and prediction method | |
| AU2020286320B2 (en) | Multi-granularity spark super trust fuzzy method applied to large-scale brain medical record segmentation | |
| CN110728294A (en) | Cross-domain image classification model construction method and device based on transfer learning | |
| CN104835181B (en) | A kind of method for tracking target based on ranking and fusing study | |
| Xu et al. | Deep neural network approximated dynamic programming for combinatorial optimization | |
| CN112364908A (en) | Decision tree-oriented longitudinal federal learning method | |
| CN112990385A (en) | Active crowdsourcing image learning method based on semi-supervised variational self-encoder | |
| CN112149990B (en) | Fuzzy supply and demand matching method based on prediction | |
| CN112380433A (en) | Recommendation meta-learning method for cold-start user | |
| US20130311410A1 (en) | Information Processing Apparatus, Information Processing Method, and Program | |
| EP4220495A1 (en) | Task learning system and method, and related device | |
| CN112948872B (en) | Road network track semantic privacy protection method based on segmented clustering | |
| US20220188649A1 (en) | Decision tree-oriented vertical federated learning method | |
| CN113609379B (en) | Label system construction method and device, electronic equipment and storage medium | |
| Davani et al. | An autonomous system for efficient control of ptz cameras | |
| Qu et al. | A dynamic ridesplitting method with potential pick-up probability based on GPS trajectories | |
| US20220351090A1 (en) | Federated learning method for decision tree-oriented horizontal | |
| US20230244700A1 (en) | System and method for identifying approximate k-nearest neighbors in web scale clustering | |
| US20220398416A1 (en) | System and method for identifying approximate k-nearest neighbors in web scale clustering | |
| CN111553374A (en) | Road scene dividing method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FGA | Letters patent sealed or granted (standard patent) |