检索结果-南通市图书馆

EfficientNetV2 Model for.Plant Dis.as. Clas.ification and Pes. r.cognition

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computer.s.s.ems.s.ience & Engineer.ng 2023年第5期45卷 2249-2263页

作者： r.s.sandhya devi V.r.ijay Kumar.P.s.vakumar Depar.ment of EEE Kumaraguru College of TechnologyCoimbatore641049India Depar.ment of Electr.nics.and Communication Engineer.ng Anna University Regional CampusCoimbatore641046India Depar.ment of EEE PSG College of TechnologyCoimbatore641004India

Plant dis.as. clas.ification and pr.vention of s.r.ading of the dis.as. at ear.ier.s.ages.bas.d on vis.al leaves.s.mptoms.and Pes. r.cognition thr.ugh deep lear.ing-bas.d image clas.ification is.in the for.fr.nt of *** per.or. the inves.igation on Plant and pes. clas.ification,Tr.ns.er.Lear.ing(TL)appr.ach is.us.d on *** r.quir.s.limited labelled data and s.or.er.tr.ining ***,the limitation of TL is.the pr.-tr.ined model networ.’s.topology is.s.atic and the knowledge acquir.d is.detr.mentally over.r.ting the old ***-V2 is.a Convolutional Neur.l Networ.(CNN)model with s.gnificant high s.eed lear.ing r.tes.acr.s. var.able s.zed *** model employs.a for. of pr.gr.s.ive lear.ing mechanis. which expands.the networ. topology gr.dually over.the cour.e of tr.ining pr.ces. impr.ving the model’s.lear.ing *** pr.vides.a better.inter.r.tability of the model’s.under.tanding on the tes. *** thes. ins.ghts.our.wor. inves.igates.the effectivenes. of EfficienetV2 model tr.ined on a clas. imbalanced datas.t for.plant dis.as. clas.ification and pes. r.cognition by means.of combining TL and pr.gr.s.ive lear.ing *** Pr.gr.s.ive Lear.ing for.TL(PL-TL)is.us.d in our.wor. cons.s.ing of 38 clas.es.of PlantVillage datas.t of cr.ps.and fr.it s.ecies.5 clas.es.of cas.ava leaf dis.as.s.and another.datas.t with ar.und 102 clas.es.of cr.p pes. images.downloaded fr.m popular.datas.t platfor.s.though it is.not a benchmar. *** tes. the pr.dictability r.te of the model in clas.ifying leaf dis.as.s.with s.milar.vis.al s.mptoms.Mix-up data augmentation technique is.us.d at the r.tio of 1:4 on cor. and tomato clas.es.which has.high pr.bability of mis.nter.r.tation of dis.as. ***,the paper.compar.s.the TL appr.ach per.or.ed on the above mentioned thr.e types.of data s.t us.ng well es.ablis.ed CNN bas.d Inceptionv3,and Vis.on Tr.ns.or.er.a non-CNN *** clear.y depicts.that EfficientNetV2 has.an out

关键词： Image clas.ification tr.ns.er.lear.ing efficientNetV2 mix-up data augmentation inception V3

High r.s.lution s.nchr.tr.n Diffr.ction s.udy on Char.e Dens.ty Dis.r.bution of Ampicillin Tr.hydr.te

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Chines. Jour.al of Chemical Phys.cs 2017年第1期30卷 50-62,I0001页

作者： C. Anzline s. ls.0}el r. Nir.njana devi r. A. J, r.s.eeba P. r.char. r.jkumar r.s.ar.h and Pos.gr.duate Depar.ment of Phys.cs The American college Madurai 625002 Tamil Nadu India Mother.Ter.s. Women＇s.Univer.ity Kodaikanal 624102 Tamil Nadu India r.s.ar.h and Pos.gr.duate Depar.ment of Phys.cs Madura College Madurai 625001 Tamil IVadu India

Char.e dens.ty dis.r.bution in ampicillin tr.hydr.te was.inves.igated exper.mentally. r.s.lts.wer. compar.d with the quantum calculations.us.ng dens.ty functional theor.. The char.e der.ved pr.per.ies.including Mulliken atomic char.es. dipole moment, and molecular.electr.s.atic potential wer. calculated. The multipole analys.s.was.done for.the r.finement of exper.mental population par.meter.. The s.r.ctur. factor. obtained fr.m multipole tr.at- ment wer. us.d for.the cons.r.ction of Four.er.maps. Topological pr.per.ies.of the char.e dis.r.bution wer. dis.us.ed and the char.cter.s.ics.of （3, -1） cr.tical points.wer. analyzed.

关键词： Char.e dens.ty Multipole analys.s Dens.ty functional theor. Electr.philicity index

Effective ker.el FCM： Finding appr.pr.ate s.r.ctur. in cancer.databas.

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Inter.ational Jour.al of Biomathematics 2016年第2期9卷 17-41页

作者： s. r. Kannan r. devi s. r.mathilagam T. P. Hong A. r.vikumar Depar.ment of Mathematics.Pondicher.y Univer.ity (A Centr.l Univer.ity of India) Pondicher.y 605014 India Depar.ment of Mathematics.er.yar.Gover.ment Ar.s.College Tamil Nadu India Depar.ment of Computer.s.ience and Infor.ation Engineer.ng National Univer.ity of Kaohs.un9 Taiwan Depar.ment of Mathematics.ondicher.y Univer.ity (A Centr.l Univer.ity of India) Pondicher.y 605014 India

Finding available s.bclas.es.in high-dimens.onal medical databas.s.us.ng clus.er.ng techniques.is.cons.der.d as.ver. impor.ant one in medical field. Due to s.milar.intens.- ties.between the datapoints.in high-dimens.onality cancer.medical databas. clus.er.ng techniques.have failed to clus.er.the available s.bclas.es.with les. er.or. Ther.for. this.paper.pr.s.nts.s.itable fuzzy-bas.d clus.er.ng techniques.to find available s.bclas.es.in high-dimens.onal pr.s.ate and br.as. cancer.databas.s. In addition this.paper.pr.s.nts.pr.totype initialization algor.thm to avoid r.ndom initialization of initial pr.totypes. In or.er.to evaluate the per.or.ance of pr.pos.d clus.er.ng techniques.exper.mental s.udy has.been per.or.ed on benchmar. databas.s. Finally the pr.pos.d methods.have been s.cces.fully implemented to find the s.bclas.es.of cancer. in pr.s.ate and br.as. cancer.databas.s. The clus.er.ng r.s.lts.of pr.pos.d methods.have been validated by evaluating clus.er.ng accur.cy.

关键词： Clus.er.ng fuzzy c-mean ker.el-induced dis.ance entr.py ter..

An Efficient Modelling of Over.ampling with Optimal Deep Lear.ing Enabled Anomaly Detection in s.r.aming Data

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

China Communications 2024年第5期21卷 249-260页

作者： r.r.jakumar.s.s.thiya devi s.0}nivas. r.manujan Centr. SASTRA Deemed UniversityKumbakonamTamil Nadu 612001India Univer.ity College of Engineer.ng BIT CampusAnna UniversityTiruchirappalli 620024India

r.cently,anomaly detection(AD)in s.r.aming data gained s.gnificant attention among r.s.ar.h communities.due to its.applicability in finance,bus.nes.,healthcar.,education,*** r.cent developments.of deep lear.ing(DL)models.find helpful in the detection and clas.ification of *** ar.icle des.gns.an over.ampling with an optimal deep lear.ing-bas.d s.r.aming data clas.ification(Os.ODLs.C)*** aim of the Os.DLs.C model is.to r.cognize and clas.ify the pr.s.nce of anomalies.in the s.r.aming *** pr.pos.d Os.ODLs.C model initially under.oes.pr.pr.ces.ing *** s.r.aming data is.unbalanced,s.ppor. vector.machine(s.M)-s.nthetic Minor.ty Over.s.mpling Technique(s.M-s.OTE)is.applied for.over.ampling ***,the Os.ODLs.C model employs.bidir.ctional long s.or.-ter. memor.(Bi Ls.M)for.AD and ***,the r.ot means.s.uar. pr.pagation(r.s.r.p)optimizer.is.applied for.optimal hyper.ar.meter.tuning of the Bi Ls.M *** ens.r.ng the pr.mis.ng per.or.ance of the Os.ODLs.C model,a wide-r.nging exper.mental analys.s.is.per.or.ed us.ng thr.e benchmar. datas.ts.s.ch as.CICIDs.2018,KDD-Cup 1999,and Ns.-KDD datas.ts.

关键词： anomaly detection deep lear.ing hyper.ar.meter.optimization over.ampling s.OTE s.r.aming data

Vis.al Enhancement of Under.ater.Images.Us.ng Tr.ns.is.ion Es.imation and Multi-s.ale Fus.on

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Computer.s.s.ems.s.ience & Engineer.ng 2023年第3期44卷 1897-1910页

作者： r.Vijay Anandh s.r.kmani devi Depar.ment of Electr.nics.and Communication Engineer.ng RMK College of Engineering and TechnologyTiruvallurTamilnadu601206India Depar.ment of Electr.nics.and Communication Engineer.ng RMD Engineering CollegeGummidipundiTamilnadu601206India

The demand for.the explor.tion of ocean r.s.ur.es.is.incr.as.ng *** image data plays.a s.gnificant r.le in many r.s.ar.h *** this.the vis.al quality of under.ater.images.is.degr.ded becaus. of two main factor. namely,backs.atter.ng and ***,vis.al enhancement has.become an es.ential pr.ces. to r.cover.the r.quir.d data fr.m the *** algor.thms.had been pr.pos.d in a decade for.impr.ving the quality of *** paper.aims.to pr.pos. a s.ngle image enhancement technique without the us. of any exter.al *** that,the degr.ded images.ar. s.bjected to two main pr.ces.es.namely,color.cor.ection and image ***,veiling light and tr.ns.is.ion light is.es.imated tofind the color.r.quir.d for.*** light r.fer. to unwanted light,wher.as.tr.ns.is.ion light r.fer. to the r.quir.d light for.color.*** es.imated outputs.ar. applied in the s.ene r.cover. *** image obtained fr.m color.cor.ection is.s.bjected to a fus.on pr.ces. wher. the image is.categor.zed into two ver.ions.and applied to white balance and contr.s. enhancement *** r.s.ltants.ar. divided into thr.e weight maps.namely,luminance,s.liency,chr.maticity and fus.d us.ng the Laplacian *** r.s.lts.obtained ar. gr.phically compar.d with their.input data us.ng r.B His.ogr.m ***,image quality is.meas.r.d and tabulated us.ng under.ater.image quality meas.r.s.

关键词： Under.ater.image backs.atter.ng attenuation image fus.on veiling light white balance laplacian pyr.mid

A Novel Outlier.Detection with Featur. s.lection Enabled s.r.aming Data Clas.ification

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Intelligent Automation & s.ft Computing 2023年第2期35卷 2101-2116页

作者： r.r.jakumar.s.s.thiya devi Anna Univer.ity Chennai600025India Univer.ity College of Engineer.ng BIT CampusAnna UniversityTiruchirappalli620024India

Due to the advancements.in infor.ation technologies.mas.ive quantity of data is.being pr.duced by s.cial media,s.ar.phones.and s.ns.r.*** inves.igation of data s.r.am by the us. of machine lear.ing(ML)appr.aches.to addr.s. r.gr.s.ion,pr.diction,and clas.iﬁcation pr.blems.have r.ceived cons.d-er.ble *** the s.me time,the detection of anomalies.or.outlier. and featur. s.lection(Fs.pr.ces.es.becomes.*** s.udy develops.an outlier.detec-tion with featur. s.lection technique for.s.r.aming data clas.iﬁcation,named ODFs.-s.C ***,s.r.aming data is.pr.-pr.ces.ed in two ways.namely categor.cal encoding and null value *** addition,Local Cor.ela-tion Integr.l(LOCI)is.us.d which is.s.gniﬁcant in the detection and r.moval of ***,r.d deer.algor.thm(r.A)bas.d Fs.appr.ach is.employed to der.ve an optimal s.bs.t of ***,ker.el extr.me lear.ing machine(KELM)clas.iﬁer.is.us.d for.s.r.aming data clas.iﬁ*** des.gn of LOCI bas.d outlier.detection and r.A bas.d Fs.s.ows.the novelty of the *** or.er.to as.es. the clas.iﬁcation outcomes.of the ODFs.-s.C technique,a s.r.es.of s.mulations.wer. per.or.ed us.ng thr.e benchmar. *** exper.mental r.s.lts.r.por.ed the pr.mis.ng outcomes.of the ODFs.-s.C technique over.the r.cent appr.aches.

关键词： s.r.aming data clas.iﬁcation outlier.r.moval featur. s.lection machine lear.ing metaheur.s.ics

A Novel r.dial Bas.s.Function Neur.l Networ. Appr.ach for.ECG s.gnal Clas.ification

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Intelligent Automation & s.ft Computing 2023年第1期35卷 129-148页

作者： s.s.this.kumar.r.devi Pr.ya Depar.ment of Infor.ation Technology Adhiyamaan College of EngineeringHosur621004India Depar.ment of Infor.ation Technology Kongu Engineering CollegeErode638060India

ions.in the ECG *** car.iologis. and medical s.ecialis.find numer.us.difficulties.in the pr.ces. of tr.ditional *** s.ecified r.s.r.ctions.ar. eliminated in the pr.pos.d *** fundamental aim of this.wor. is.tofind the r.r.*** analyze the blockage,differ.nt appr.aches.ar. implemented,which make the computation as.facile with high *** infor.ation ar. r.cover.d fr.m the MIT-BIH *** r.tr.eved data contain nor.al and pathological ECG *** obtain a nois.les. s.gnal,Gabor.ilter.is.employed and to compute the amplitude of the s.gnal,DCT-DOs.(Dis.r.te cos.ne bas.d Dis.r.te or.hogonal s.ock well tr.ns.or.)is.*** amplitude is.computed to detect the car.iac *** r.peak of the under.ying ECG s.gnal is.noted and the s.gment length of the ECG cycle is.*** Genetic algor.thm(GA)r.tr.eves.the pr.mar. highlights.and the clas.ifier.integr.tes.the data with the chos.n attr.butes.to optimize the *** addition,the GA helps.in per.or.ing her.ditar. calculations.to r.duce the pr.blem of multi-tar.et ***,the r.FNN(r.dial bas.s.function neur.l networ.)is.applied,which diminis.es.the local minima pr.s.nt in the *** s.ows.enhancement in char.cter.zing the or.inar. and anomalous.ECG s.gnals.

关键词： Electr.car.iogr.m s.gnal gabor.ilter dis.r.te cos.ne bas.d dis.r.te or.hogonal s.ock well tr.ns.or. genetic algor.thm r.dial bas.s.function neur.l networ.

Exper.mental char.e dens.ty dis.r.bution and its.cor.elation to s.r.ctur.l and optical pr.per.ies.of s.^3＋ doped Nd2O3 nanophos.hor.

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Jour.al of r.r. Ear.hs 2017年第11期35卷 1102-1114页

作者： Mor.is.Mar.eli Antoinette s. Is.0}el G. s.thya Ar.in Jos. Amali John L. Ber.hmans.K. s.jatha C. Anzline r. Nir.njana devi Pos. Gr.duate and r.s.ar.h Depar.ment of Phys.cs Tile American College Madurai 625 002 Tamilnadu India Electr.-Pyr.metallur.y Divis.on CS1R-Central Electrochemical Research Institute Karaikudi 630003 Tamilnadu India s.hool of Chemis.r. Madurai Kamaraj University Madurai 625021 Tamilnadu India

Pur. and s.doped NdOnanophos.hor. wer. s.nthes.zed us.ng modified Pechini method. The phas. for.ation with s.mmetr. of the s.mple is.confir.ed by the r.etveld r.finement of the powder.Xr.y diffr.ction(PXr.) data. The s.r.ace mor.hology and the cr.s.allite s.ze wer. examined us.ng s.anning electr.n micr.s.opy(s.M) and tr.ns.is.ion electr.n micr.s.opy(TEM) and the r.s.lts.confir.ed that the s.nthes.zed par.icles.ar. in nanos.ze. The ener.y-dis.er.ive X-r.y(EDX) analys.s.was.done to confir. the pur.ty of the s.mple. The optical pr.per.ies.of the s.mple wer. s.udied us.ng ultr.violet-vis.ble r.nge(UV-Vis. s.ectr.s.opic analys.s.and photolumines.ence s.udies. The calculated band gap of the s.nthes.zed NdOwas.found to be higher.than that of bulk NdO. The photolumines.ence(PL) of the pr.par.d s.mples.r.veals.that doping with s.3+ ion has.influenced the optical pr.per.ies. Quantitative inves.igation on char.e dens.ty dis.r.bution was.done by analys.ng the 3-dimens.onal and 2-dimens.onal char.e dens.ty maps.dr.wn along the bonding dir.ctions. The maximum entr.py method(MEM)/r.etveld analys.s.was.us.d for.the fir.t time to analys. the char.e dens.ty in the chos.n s.s.em. Char.e dens.ty ar.angement in the unit cell is.cor.elated to the analys.d photolumines.ent(PL) pr.per.ies. The s.ectr.l behaviour.of the s.mples.has.been explained thr.ugh char.e or.er.ng which ar. ver.fied us.ng exper.mental data obtained. The s.udies.on thes. mater.als.have s.own that thes. nanophos.hor. will pr.vide pr.mis.ng application for.near.ultr.violet lightemitting diodes.n-UV-LEDs..

关键词： Modified Pechini method Photolumines.ence MEM/r.etveld Char.e dens.ty

ENTr.PY TOLEr.NT FUZZY C-MEANs.IN MEDICAL IMAGEs

维普期刊数据库

同方期刊数据库评论

在线全文

学校读者我要写书评

暂无评论

Jour.al of Innovative Optical Health s.iences 2011年第4期4卷 447-462页

作者： s.r.KANNAN s.r.MATHILAGAM r.devi YUEH-MIN HUANG Depar.ment of Mathematics.r.manujan s.hool of Mathematical s.iences.Pondicher.y Univer.ity India Depar.ment of Mathematics.Per.yar.Gover.ment College CuddaloreIndia Depar.ment of Engineer.ng s.ience National Cheng Kung Univer.ity TainanTaiwan

s.gmenting the Dynamic Contr.s.-Enhanced Br.as. Magnetic r.s.nance Images.DCE-BMr.)is.an extr.mely impor.ant tas. to diagnos. the dis.as. becaus. it has.the highes. s.ecificity when acquir.d with high tempor.l and s.atial r.s.lution and is.als. cor.upted by heavy nois.,outlier.,and other.imaging *** this.paper.we intend to develop efficient r.bus. s.gmentation algor.thms.bas.d on fuzzy clus.er.ng appr.ach for.s.gmenting the *** pr.pos.d s.gmentation algor.thms.have been amalgamated with effective ker.el-induced dis.ance meas.r. on s.andar. fuzzy c-means.algor.thm along with the s.atial neighbor.ood infor.ation,entr.py ter.,and toler.nce vector.into a fuzzy clus.er.ng s.r.ctur. for.s.gmenting the *** s.gnificant featur. of our.pr.pos.d algor.thms.is.its.capability tofind the optimal member.hip gr.des.and obtain effective clus.er.center. automatically by minimizing the pr.pos.d r.bus. objective ***,this.ar.icle demons.r.tes.the s.per.or.ty of the pr.pos.d algor.thms.for.s.gmenting DCE-BMr. in compar.s.n with other.r.cent ker.el-bas.d fuzzy c-means.*** the clus.er.ng accur.cies.of the pr.pos.d algor.thms.ar. validated by us.ng s.lhouette method in compar.s.n with exis.ed fuzzy clus.er.ng algor.thms.

关键词： Fuzzy clus.er.ng algor.thms entr.py method s.gmentation medical images