Robust Low-Rank Regularized Regression for Face Recognition with Occlusion
|
|
- Dominick Lawrence
- 5 years ago
- Views:
Transcription
1 Robust Low-Rank Regularzed Regresson for ace Recognton wth Occluson Janjun Qan, Jan Yang, anlong Zhang and Zhouchen Ln School of Computer Scence and ngneerng, Nanjng Unversty of Scence and echnology Key Lab. of Machne Percepton (MO), Pekng Unversty Abstract Recently, regresson analyss based classfcaton methods are popular for robust face recognton. hese methods use a pel-based error model, whch assumes that errors of pels are ndependent. hs assumpton does not hold n the case of contguous occluson, where the errors are spatally correlated. Observng that occluson n a face mage generally leads to a low-rank error mage, we propose a low-rank regularzed regresson model and use the alternatng drecton method of multplers (ADMM) to solve t. We thus ntroduce a novel robust low-rank regularzed regresson (RLR 3 ) method for face recognton wth occluson. Compared wth the estng structured sparse error codng models, whch perform error detecton and error support separately, our method ntegrates error detecton and error support nto one regresson model. perments on benchmark face databases demonstrate the effectveness and robustness of our method, whch outperforms state-of-the-art methods.. Introducton Automatc face recognton has been a hot topc n the area of computer vson and pattern recognton due to the ncreasng need for real-world applcatons []. Recently regresson analyss becomes a popular tool for face recognton. Naseem et al. presented a lnear regresson classfer (LRC) for face classfcaton [3]. Wrght et al. proposed a sparse representaton based classfcaton (SRC) method to dentfy human faces wth varyng llumnaton changes, occluson and real dsguse []. A test sample mage s coded as a sparse lnear combnaton of the tranng mages, and then the classfcaton s made by dentfyng whch class yelds the least reconstructon resdual. Although SRC performs well n face recognton, t lacks theoretcal justfcaton. Yang et al. gave an nsght nto SRC and sought reasonable supports for ts effectveness [3]. hey thought that the L -optmzer has two propertes, sparseness and closeness. Sparseness determnes a small number of nonzero representaton coeffcents and closeness makes the nonzero representaton coeffcents concentrate on the tranng samples havng the same class label as the test sample. However, the L 0 -optmzer can only acheve sparseness. Yang et al. constructed a Gabor occluson dctonary to mprove the performance and effcency of SRC [4]. Yang and Zhang proposed a robust sparse codng (RSC) model for face recognton [6]. RSC s robust to varous knds of outlers (e.g., occluson and facal epresson). Based on the mamum correntropy crteron, He et al. [7, 8] presented robust sparse representaton for face recognton. Recently, some researchers have begun to queston the role of sparseness n face recognton [9, 0]. In [], Zhang et al. analyzed the work rule of SRC and beleved that t s the collaboratve representaton that mproves the classfcaton performance, rather than the L -norm sparseness. hey further ntroduced the collaboratve representaton based classfcaton (CRC) wth the non-sparse L -norm to regularze the representaton coeffcents. CRC can acheve smlar results as SRC and sgnfcantly speed up the algorthm. he regresson methods mentoned above all use the pel-based error model [6], whch assumes that errors of pels are ndependent. hs assumpton does not hold n the case of contguous occluson, where errors are spatally correlated [5]. In addton, characterzng the representaton error pel by pel ndvdually neglects the whole structure of the error mage. o address these problems, Zhou et al. ncorporated the Markov Random eld model nto the sparse representaton framework for spatal contnuty of the occluson [5]. L et al. eplored the ntrnsc structure of contguous occluson and proposed a structured sparse error codng (SSC) model []. hese two works share the same two-step teraton strategy: () Detectng errors va sparse representaton or codng, and () stmatng error supports (.e. determnng the real occluded part) usng graph cuts. he dfference s that SSC uses more elaborate technques, such as the teratvely reweghted sparse codng n the error detecton step and a morphologcal graph model n the error support step for achevng better performance. However, SSC does not numercally converge to the desred soluton; t needs an addtonal qualty assessment model to choose the desred soluton from the teraton sequence. 43
2 Low-rank Orgnal Image Image wth Occluson Some recent works pont out that the vsual data has lowrank structure [7]. Most of the etng methods am to fnd a low-rank appromaton for matr completon. However, the rank mnmzaton problem s NP hard n general. In [4], azel et al. appled the nuclear norm heurstc to solve the rank mnmzaton problem, where the nuclear norm of a matr s the sum of ts sngular values. Based on these results, robust prncple component analyss (RPCA) s proposed to decompose an mage nto two parts: data matr (low-rank) and the nose (sparse) [5, 6]. Zhang et al. ntroduced a matr completon algorthm based on the runcated Nuclear Norm Regularzaton for estmatng mss values [7]. Ma et al. ntegrated rank mnmzaton nto sparse representaton for dctonary learnng and appled the model for face recognton [8]. Chen et al. presented a novel low-rank matr appromaton algorthm wth structural ncoherence for robust face recognton [9]. hs paper focuses on face recognton wth occluson. We observe that contguous occluson n a face mage generally leads to a low-rank representaton error mage (.e. error mage or nose mage), as shown n g.. Motvated by ths observaton, we add a rank functon (whch s replaced by a nuclear norm for easy optmzaton) of the representaton error mage nto a regresson model. he model can be solved va the alternatng drecton method of multplers (ADMM) []. he proposed method has the followng merts: () Compared wth state-of-the-art regresson methods such as SRC, RSC and CSR, whch characterze the representaton error ndvdually and neglect the whole structure of the error mage, our model vews the error mage as a whole and takes full use of ts low-rank structure. () Compared wth SSC [] and Zhou s method [5], whch perform the error detecton step and the error support step teratvely but cannot guarantee the convergence of the whole algorthm, our method ntegrates error detecton and error support nto one regresson model, and ts ADMM algorthm theoretcally converges well.. Low-Rank Regularzed Regresson In ths secton, we present the low-rank regularzed regresson model to code the mage and use the alternatng drecton method of multplers [] to solve the model... Low-Rank Regularzed Regresson Suppose that we are gven a dataset of n matrces,, d l A An R d l and a matr Y R. Let us represent Y lnearly by takng the followng form: Y ( ), () where ( ) A A nan,,, n s the representaton coeffcent vector. s the nose (representaton error). Generally, the can be determned by solvng the followng optmzaton problem (lnear regresson): where mn ( ) Y, () s the robenus norm of a matr. o avod over fttng, we often solve the followng regularzed model (rdge regresson) nstead: mn ( ) Y. (3) he above optmzaton problem can be solved n a closed form. he rank of a matr s a good tool to descrbe the structural characterstcs of an error mage. he error (representaton error) mage, as shown n g., s typcally low rank as opposed to the full rank orgnal mage. However, the estng lnear regresson models do not make use of ths knd of structural nformaton. o address ths problem, we ntroduce the low-rank regularzaton to the rdge regresson model. Specfcally, the optmzaton problem s formulated as follows: mn ( ) Y rank(( ) Y), (4) where s the balance factor and s the regularzed parameter. he optmzaton problem of q. (4) s etremely dffcult to solve due to the dscrete nature of the rank functon. ortunately, as suggested by the matr completon method [4, 0, ], the rank mnmzaton problem can be replaced by the nuclear norm mnmzaton problem: mn ( ) Y ( ) Y. (5) In the followng secton, we wll develop the optmzaton algorthm to solve q. (5) by usng the alternatng drecton method of multplers []. Resdual Image gure : he mage wth block occluson s lnearly represented by s dfferent mages and the resdual (nose) mage. 43
3 .. Optmzaton va ADMM In ths secton, we show how the alternatng drecton method of multplers (ADMM) can be adopted to solve q. (5) effcently. or more detals of ADMM, we refer readers to []. o deal wth our problem, the model n q. (5) s rewrtten as: mn,, (6) st.. ( ) Y. he augmented Lagrange functon s gven by: (,, ) L Z + where 0 r Z ( ) Y ( ) Y, s a penalty parameter, Z s the Lagrange multpler, and r( ) s the trace operator. ADMM conssts of the followng teratons: k arg mn ( ) (8) (7) L k arg mn L ( ) (9) k k k k Z Z (( ) Y ) (0) Updatng Denote H [Vec( A),,Vec( An)], g Vec( + Y Z ), where Vec( ) convert matr nto a vector, then the objectve functon L ( ) n q. (8) s equvalent to L ( ) r Z ( ) ( ) Y () ( ) ( Y Z) r ( Y) Z ZZ. he problem of q. (8) can be reformulated as: k arg mn H g. () q. () s actually a rdge regresson model. So we can obtan the soluton of q. () by: ( H H I) H g (3) k Updatng he objectve functon ( ) n q. (9) can be rewrtten by L L ( ) r Z ( ) Y (4) Y Z r Z Y Y r (( ) ) ( ) r ( ) const Algorthm Solvng LR 3 va ADMM Input: A set of matrces A,,A and a n p q matr Y R, the model parameter, the termnaton condton parameter. 0 0 Intalze, Z, whle do ( ) Y Z const where const and const are constant terms, whch are ndependent of the varable. he optmzaton problem q. (9) can be reformulated as k arg mn ( ) Y Z (5) As suggested by [3], the above optmzaton problem s solved by k U S V [ ], where ( U, S, V ) = svd ( ) Y Z. he sngular value shrnkage operator (6) [ S] s defned as [ S ] dag {ma(0, S j j )} j n. (7), Stoppng crteron As suggested n [], the stoppng crteron of the algorthm s: the prmal resdual must be small: k k ( ) Y, and the dfference between successve teratons should also be small: k k k k ma,, where s a gven k k ( ) Y or k k k k ma, ( H H I) H g, k k arg mn Y tolerance. In summary, the pseudo code of our method to solve q. (6) s shown n Algorthm. Algorthm can be nterpreted as usng the two-step teraton strategy for robust face recognton as those used n [5, ]. he step of updatng s actually an error detecton ( ) Y Z, k k k k Z Z Y (( ) ). end whle Output: Optmal representaton coeffcent. 3433
4 step for determnng the representaton coeffcents and representaton errors, and the step of updatng s actually an error support step for determnng the real occluded part. So we can say that LR 3 provdes a unfed framework to ntegrate error detecton and error support detecton nto one smple model. 3. Classfcaton based on Robust Low-Rank Regularzed Regresson We notce that SSC [] adopts a robust sparse representaton model,.e. teratvely reweghted sparse codng n the error detecton step, but our low-rank regularzed regresson (LR 3 ) only uses a smple rdge regresson model for updatng. o further mprove the robustness of the proposed method, n ths secton we borrow the dea of robust regresson to our model and ntroduce a robust low-rank regularzed regresson model (RLR 3 ): mn W(( ) Y) W (( ) Y). (8) where W s a weght matr of the representaton error, and denotes the Hadamard product of two matrces. he robust low-rank regularzed regresson model can be solved by usng the teratvely reweghted process [6]. ach teraton step s to solve a low-rank regularzed regresson problem. Specfcally, gven a test sample Y, we compute the representaton coeffcent va Algorthm and the representaton error of Y n order to ntalze the weght. he representaton error s ntalzed as Y Y n, where Y n s the ntal estmaton of the mages from the gallery set. In ths study, we smply set Y n as the mean mage of all samples n the codng dctonary snce we do not know whch class the test mage Y belongs to. Wth the ntalzed Y n, our method can estmate the weght matr W teratvely. W, s the weght assgned to each pel of the test mage. he weght functon [6] s: W ep( (, j) ), j, j, (9) ep( ( ) ) where and are postve scalars. he convergence s acheved when the dfference between the weghts n successve teratons satsfes the followng condton: ( t) ( t) ( t) W W / W. (0) Based on the optmzaton soluton va the teratve process, we obtan a weghted dctonary B [ B Bn], where B W D,,, n and D s the codng dctonary whch s composed of the tranng samples. he test sample Y s reconstructed as Yˆ jb, j, where j ( ) ( ) s the functon that selects the ndces of the coeffcents assocated wth the -th class. he correspondng reconstructon error of the -th class s defned as: r ( y) W ( Y Yˆ ). () he decson rule s: f r ( y) mn r ( y ), then y s assgned to Class l. able he recognton rates (%) of each classfer for face recognton on the AR database wth dsguse occluson. Methods Sunglasses Scarves CRC LR SRC [] CSR [8] SSC RSC [6] RLR perments In ths secton, we compare the proposed methods LR 3 and RLR 3 wth CRC, SRC, CSR, SSC and RSC. In our eperments, there are fve parameters of the proposed RLR 3. he parameters and n q. (9) follows the suggeston n [6]. he default value for penalty parameter s. Both the balance factor and the regularzed parameter are ntroduced n the followng eperments. 4.. ace Recognton wth Real Dsguse In our eperments, we only use a subset of AR face mage database [4]. he subset contans 00 ndvduals, 50 males and 50 females. All the ndvduals have two sesson mages and each sesson contans 3 mages. he face porton of each mage s manually cropped and then normalzed to 4 30 pels. he frst eperment chooses the frst four mages (wth varous facal epressons) from sesson and sesson of each ndvdual to form the tranng set. he total tranng mages s 800. here are two test sets: the mages wth sunglasses and the mages wth scarves. ach set contans 00 mages (one mage per sesson of each ndvdual wth neutral epresson). he balance factor s 0 and 0 - for the test mages wth sunglasses and scarves, respectvely. he regularzed parameter s able lsts the recognton rates of CRC, SRC, CSR, SSC, RSC, LR 3 and RLR 3. rom able, we can see that RLR 3 acheves the best performance among all the methods. LR 3 also gves better results than CRC. Both RSC and CSR obtan the same results as RLR 3 when the test mages are wth sunglasses. However, the results of SRC and CSR are l 4434
5 sgnfcantly lower than those of RLR 3 when the test mages are wth scarves. In the second eperment, four neutral mages wth dfferent llumnaton from the frst sesson of each ndvdual are used for tranng. he dsguse mages wth varous llumnaton and glasses or scarves per ndvdual n sesson and sesson for testng. he balance factor s 0 - and the regularzaton parameter s he recognton rates of each method are lsted n able. rom able, we can see that RLR 3 sgnfcantly outperforms CRC, LR 3, SRC, CSR, SSC and RSC on dfferent test subsets. SRC and CSR perform well on mages wth sunglasses and poorly on mages wth scarves. SSC gves smlar results as RSC n dfferent cases. Compared to RSC, 3.0%,.0%, 4.0% and 4.6% mprovement are acheved by RLR 3 on four dfferent testng sets. (a) able he recognton rates (%) of each classfer for face recognton on the AR database wth dsguse occluson. Methods Sunglasses Scarves Sesson Sesson Sesson Sesson CRC LR SRC [] CSR [8] SSC RSC [6] RLR ace Recognton wth Random Block Occluson he etended Yale B face mage database [5] contans 38 human subjects under 9 poses and 64 llumnaton condtons. All frontal-face mages marked wth P00 are used n our eperment, and each s reszed to pels. In the frst eperment, we use the same eperment settng as n [] to test the robustness of RLR 3. Subsets and of tended Yale B are used for tranng and subset 3 wth the unrelated block mages s used for testng. Both and are set to 0. g. (a) plots recognton rates of CRC, SRC, CSR, SSC, RSC, LR 3 and RLR 3 under dfferent levels of occlusons (from 0% to 50%). Wth the ncrement of the level of occluson, RLR 3 begns to sgnfcantly outperform the other methods. When the occluson percentage s 50%, the recognton rate of RLR 3 s 0.4%,.6%, 36.9% and 9% hgher than RSC, SSC, CSR and SRC, respectvely. he second eperment settng s smlar to that of the frst eperment. he only dfference s that subset 3 wth nose block mages s used for testng. s 0. and the regularzaton parameter s set to0. he recognton rates of each method versus the varous levels of occluson (from (b) (c) gure : he recognton rates (%) of CRC, SRC, CSR, SSC, RSC, LR 3 and RLR 3 wth the occluson percentage beng from 0% to 50%. (a) the test mages are wth unrelated block mages; (b) the test mages are wth nose block mages; (c) the test mages are wth mture nose. 5435
6 0% to 50%) are shown n g. (b). rom g. (b), we observe that the proposed RLR 3 sgnfcantly outperforms CRC, SRC, CSR, SSC and RSC. he performances of SRC and CSR are not good n ths case. SSC gves good performance when the occluson level s hgher. However, SSC cannot perform well when the occluson level s lower. RSC acheves comparable results when the occluson percentage s lower than 40%. However, the recognton rate of RLR 3 s 6.% hgher than that of RSC when the occluson percentage s 50%. In the thrd eperment, subsets and of tended Yale B are used for tranng and subset 3 wth the mture nose (pel corrupton and block occluson) s used for testng. s and the regularzaton parameter s set to0. he recognton rates of each method wth dfferent level of pel corrupton (and occluson) are shown n g. (c). Although the performance of each method degrades wth the ncrement of the mture nose level, RLR 3 stll acheves the best results among all the methods. he recognton rates of SSC are poor when facng wth the mture noses (pel corrupton and mage occluson). A probable reason s that SSC manly addresses the contguous occluson problem. 5. Conclusons In ths paper, we present a novel low-rank regularzed regresson model and apply the alternatng drecton method of multplers to solve t. he robust low-rank regularzed regresson based classfcaton (RLR 3 ) method s ntroduced for face recognton. RLR 3 takes advantage of the structural characterstcs of nose and provdes a unfed framework for ntegratng error detecton and error support nto one regresson model. tensve eperments demonstrate that the proposed RLR 3 s robust to corruptons: real dsguse and random block occluson, and yelds better performances as compared to state-of-the-art methods. References [] W. Zhao, R. Chellappa, P. J. Phllps et al., ace recognton: A lterature survey, ACM Computng Surveys, vol. 35, no. 4, pp , Dec, 003. [] J. Wrght, A. Y. Yang, A. Ganesh, S. S. Sastry, and Y. Ma. Robust face recognton va sparse representaton. I. PAMI, 3():0 7, 009. [3] J. Yang, L. Zhang, Y. Xu, and J.Y. Yang, Beyond sparsty: the role of L-optmzer n pattern classfcaton, Pattern Recognton, 45 (0) [4] M. Yang and L. Zhang. Gabor feature based sparse representaton for face recognton wth Gabor occluson dctonary. In CCV, 00. [5] Z. Zhou, A. Wagner, H. Mobah, J. Wrght, and Y. Ma. ace recognton wth contguous occluson usng Markov random felds. In ICCV, 009. [6] M. Yang, L. Zhang, J. Yang, and D. Zhang, Robust sparse codng for face recognton, In CVPR, 0. [7] R. He, W.S. Zheng, B.G. Hu, and X.W. Kong, A regularzed correntropy framework for robust pattern recognton, Neural Computaton, vol. 3, pp , 0. [8] R. He, W.S. Zheng, and B.G. Hu, Mamum correntropy crteron for robust face recognton, I. PAMI, vol. 33, no. 8, pp , 0. [9] R. Rgamont, M. Brown, and V. Lepett. Are sparse representatons really relevant for mage classfcaton? In CVPR, 0. [0] Q. Sh, A. rksson, A. Hengel, and C. Shen. Is face recognton really a compressve sensng problem? In CVPR, 0. [] L. Zhang, M. Yang, and X. C. eng. Sparse representaton or collaboratve representaton whch helps face recognton? In ICCV, 0. [] X.-X. L, D.-Q. Da, X.-. Zhang, and C.-X. Ren, Structured sparse error codng for face recognton wth occluson, I. IP, 03, (5), [3] I. Naseem, R. ogner, and M. Bennamoun. Lnear regresson for face recognton. I. PAMI, 3():06-, 00. [4] M. azel, H. Hnd, and S. Boyd, A rank mnmzaton heurstc wth applcaton to mnmum order system appromaton, n Proceedngs of the Amercan Control Conference, vol. 6, 00, pp [5]. Candès, X.D. L, Y. Ma, and J. Wrght. Robust prncpal component analyss? Journal of the ACM, 58(3): Artcle, 0. [6] J. Wrght, A. Ganesh, S. Rao, Y. Peng, and Y. Ma. Robust prncpal component analyss: act recovery of corrupted low-rank matrces va conve optmzaton, In NIPS, 009. [7] D. Zhang, Y. Hu, J. P. Ye, X. L. L, and X.. He. Matr completon by truncated nuclear norm regularzaton. In CVPR, 0. [8] L. Ma, C. Wang, B. Xao, and W. Zhou. Sparse representaton for face recognton based on dscrmnatve low-rank dctonary learnng. In CVPR, 0. [9] C.-. Chen, C.-P. We, and Y,-C, rank Wang. Low-rank matr recovery wth structural ncoherence for robust face recognton. In CVPR, 0. [0]. J. Candès and Y. Plan. Matr completon wth nose. Proceedngs of the I, 98(6):95 936, 009. []. J. Candès and B. Recht, act matr completon va conve optmzaton, Commun. ACM 55(6): -9 (0). [] S. Boyd, N. Parkh,. Chu, B. Peleato, and J. cksten. Dstrbuted optmzaton and statstcal learnng va the alternatng drecton method of multplers. oundatons and rends n Machne Learnng, 3():-, 0. [3] J.. Ca,.J. Candès, Z. Shen, A sngular value thresholdng algorthm for matr completon, SIAM Journal on Optmzaton, 00. [4] A. Martnez and R. benavente. he AR face database. echncal Report 4, CVC, 998. [5] K.C. Lee and J. Ho, and D. Dregman, Acqurng lnear subspaces for face recognton under varable lghtng, I. PAMI, 005, 7(5), pp
General Regression and Representation Model for Face Recognition
013 IEEE Conference on Computer Vson and Pattern Recognton Workshops General Regresson and Representaton Model for Face Recognton Janjun Qan, Jan Yang School of Computer Scence and Engneerng Nanjng Unversty
More informationDiscriminative Dictionary Learning with Pairwise Constraints
Dscrmnatve Dctonary Learnng wth Parwse Constrants Humn Guo Zhuoln Jang LARRY S. DAVIS UNIVERSITY OF MARYLAND Nov. 6 th, Outlne Introducton/motvaton Dctonary Learnng Dscrmnatve Dctonary Learnng wth Parwse
More informationCollaboratively Regularized Nearest Points for Set Based Recognition
Academc Center for Computng and Meda Studes, Kyoto Unversty Collaboratvely Regularzed Nearest Ponts for Set Based Recognton Yang Wu, Mchhko Mnoh, Masayuk Mukunok Kyoto Unversty 9/1/013 BMVC 013 @ Brstol,
More informationCompetitive Sparse Representation Classification for Face Recognition
Vol. 6, No. 8, 05 Compettve Sparse Representaton Classfcaton for Face Recognton Yng Lu Chongqng Key Laboratory of Computatonal Intellgence Chongqng Unversty of Posts and elecommuncatons Chongqng, Chna
More informationRobust Kernel Representation with Statistical Local Features. for Face Recognition
Robust Kernel Representaton wth Statstcal Local Features for Face Recognton Meng Yang, Student Member, IEEE, Le Zhang 1, Member, IEEE Smon C. K. Shu, Member, IEEE, and Davd Zhang, Fellow, IEEE Dept. of
More informationRobust Dictionary Learning with Capped l 1 -Norm
Proceedngs of the Twenty-Fourth Internatonal Jont Conference on Artfcal Intellgence (IJCAI 205) Robust Dctonary Learnng wth Capped l -Norm Wenhao Jang, Fepng Ne, Heng Huang Unversty of Texas at Arlngton
More informationNeurocomputing 101 (2013) Contents lists available at SciVerse ScienceDirect. Neurocomputing
Neurocomputng (23) 4 5 Contents lsts avalable at ScVerse ScenceDrect Neurocomputng journal homepage: www.elsever.com/locate/neucom Localty constraned representaton based classfcaton wth spatal pyramd patches
More informationFeature Reduction and Selection
Feature Reducton and Selecton Dr. Shuang LIANG School of Software Engneerng TongJ Unversty Fall, 2012 Today s Topcs Introducton Problems of Dmensonalty Feature Reducton Statstc methods Prncpal Components
More informationSupport Vector Machines
/9/207 MIST.6060 Busness Intellgence and Data Mnng What are Support Vector Machnes? Support Vector Machnes Support Vector Machnes (SVMs) are supervsed learnng technques that analyze data and recognze patterns.
More informationKernel Collaborative Representation Classification Based on Adaptive Dictionary Learning
Internatonal Journal of Intellgent Informaton Systems 2018; 7(2): 15-22 http://www.scencepublshnggroup.com/j/js do: 10.11648/j.js.20180702.11 ISSN: 2328-7675 (Prnt); ISSN: 2328-7683 (Onlne) Kernel Collaboratve
More informationSupport Vector Machines. CS534 - Machine Learning
Support Vector Machnes CS534 - Machne Learnng Perceptron Revsted: Lnear Separators Bnar classfcaton can be veed as the task of separatng classes n feature space: b > 0 b 0 b < 0 f() sgn( b) Lnear Separators
More informationParallelism for Nested Loops with Non-uniform and Flow Dependences
Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr
More informationLearning a Class-Specific Dictionary for Facial Expression Recognition
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 16, No 4 Sofa 016 Prnt ISSN: 1311-970; Onlne ISSN: 1314-4081 DOI: 10.1515/cat-016-0067 Learnng a Class-Specfc Dctonary for
More informationLearning the Kernel Parameters in Kernel Minimum Distance Classifier
Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department
More informationFace Recognition using 3D Directional Corner Points
2014 22nd Internatonal Conference on Pattern Recognton Face Recognton usng 3D Drectonal Corner Ponts Xun Yu, Yongsheng Gao School of Engneerng Grffth Unversty Nathan, QLD, Australa xun.yu@grffthun.edu.au,
More informationX- Chart Using ANOM Approach
ISSN 1684-8403 Journal of Statstcs Volume 17, 010, pp. 3-3 Abstract X- Chart Usng ANOM Approach Gullapall Chakravarth 1 and Chaluvad Venkateswara Rao Control lmts for ndvdual measurements (X) chart are
More informationTone-Aware Sparse Representation for Face Recognition
Tone-Aware Sparse Representaton for Face Recognton Lngfeng Wang, Huayu Wu and Chunhong Pan Abstract It s stll a very challengng task to recognze a face n a real world scenaro, snce the face may be corrupted
More informationNUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS
ARPN Journal of Engneerng and Appled Scences 006-017 Asan Research Publshng Network (ARPN). All rghts reserved. NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS Igor Grgoryev, Svetlana
More informationEYE CENTER LOCALIZATION ON A FACIAL IMAGE BASED ON MULTI-BLOCK LOCAL BINARY PATTERNS
P.G. Demdov Yaroslavl State Unversty Anatoly Ntn, Vladmr Khryashchev, Olga Stepanova, Igor Kostern EYE CENTER LOCALIZATION ON A FACIAL IMAGE BASED ON MULTI-BLOCK LOCAL BINARY PATTERNS Yaroslavl, 2015 Eye
More informationAn Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices
Internatonal Mathematcal Forum, Vol 7, 2012, no 52, 2549-2554 An Applcaton of the Dulmage-Mendelsohn Decomposton to Sparse Null Space Bases of Full Row Rank Matrces Mostafa Khorramzadeh Department of Mathematcal
More informationSupport Vector Machines
Support Vector Machnes Decson surface s a hyperplane (lne n 2D) n feature space (smlar to the Perceptron) Arguably, the most mportant recent dscovery n machne learnng In a nutshell: map the data to a predetermned
More informationRecognizing Faces. Outline
Recognzng Faces Drk Colbry Outlne Introducton and Motvaton Defnng a feature vector Prncpal Component Analyss Lnear Dscrmnate Analyss !"" #$""% http://www.nfotech.oulu.f/annual/2004 + &'()*) '+)* 2 ! &
More informationClassification / Regression Support Vector Machines
Classfcaton / Regresson Support Vector Machnes Jeff Howbert Introducton to Machne Learnng Wnter 04 Topcs SVM classfers for lnearly separable classes SVM classfers for non-lnearly separable classes SVM
More informationDetermining the Optimal Bandwidth Based on Multi-criterion Fusion
Proceedngs of 01 4th Internatonal Conference on Machne Learnng and Computng IPCSIT vol. 5 (01) (01) IACSIT Press, Sngapore Determnng the Optmal Bandwdth Based on Mult-crteron Fuson Ha-L Lang 1+, Xan-Mn
More informationHigh Dimensional Data Clustering
Hgh Dmensonal Data Clusterng Charles Bouveyron 1,2, Stéphane Grard 1, and Cordela Schmd 2 1 LMC-IMAG, BP 53, Unversté Grenoble 1, 38041 Grenoble Cede 9, France charles.bouveyron@mag.fr, stephane.grard@mag.fr
More informationOutline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1
4/14/011 Outlne Dscrmnatve classfers for mage recognton Wednesday, Aprl 13 Krsten Grauman UT-Austn Last tme: wndow-based generc obect detecton basc ppelne face detecton wth boostng as case study Today:
More informationFace Recognition University at Buffalo CSE666 Lecture Slides Resources:
Face Recognton Unversty at Buffalo CSE666 Lecture Sldes Resources: http://www.face-rec.org/algorthms/ Overvew of face recognton algorthms Correlaton - Pxel based correspondence between two face mages Structural
More informationComputer Aided Drafting, Design and Manufacturing Volume 25, Number 2, June 2015, Page 14
Computer Aded Draftng, Desgn and Manufacturng Volume 5, Number, June 015, Page 14 CADDM Face Recognton Algorthm Fusng Monogenc Bnary Codng and Collaboratve Representaton FU Yu-xan, PENG Lang-yu College
More informationS1 Note. Basis functions.
S1 Note. Bass functons. Contents Types of bass functons...1 The Fourer bass...2 B-splne bass...3 Power and type I error rates wth dfferent numbers of bass functons...4 Table S1. Smulaton results of type
More informationReview of approximation techniques
CHAPTER 2 Revew of appromaton technques 2. Introducton Optmzaton problems n engneerng desgn are characterzed by the followng assocated features: the objectve functon and constrants are mplct functons evaluated
More informationCS434a/541a: Pattern Recognition Prof. Olga Veksler. Lecture 15
CS434a/541a: Pattern Recognton Prof. Olga Veksler Lecture 15 Today New Topc: Unsupervsed Learnng Supervsed vs. unsupervsed learnng Unsupervsed learnng Net Tme: parametrc unsupervsed learnng Today: nonparametrc
More informationCluster Analysis of Electrical Behavior
Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School
More informationA Fast Content-Based Multimedia Retrieval Technique Using Compressed Data
A Fast Content-Based Multmeda Retreval Technque Usng Compressed Data Borko Furht and Pornvt Saksobhavvat NSF Multmeda Laboratory Florda Atlantc Unversty, Boca Raton, Florda 3343 ABSTRACT In ths paper,
More informationA Robust LS-SVM Regression
PROCEEDIGS OF WORLD ACADEMY OF SCIECE, EGIEERIG AD ECHOLOGY VOLUME 7 AUGUS 5 ISS 37- A Robust LS-SVM Regresson József Valyon, and Gábor Horváth Abstract In comparson to the orgnal SVM, whch nvolves a quadratc
More informationFeature Extraction Based on Maximum Nearest Subspace Margin Criterion
Neural Process Lett DOI 10.7/s11063-012-9252-y Feature Extracton Based on Maxmum Nearest Subspace Margn Crteron Y Chen Zhenzhen L Zhong Jn Sprnger Scence+Busness Meda New York 2012 Abstract Based on the
More informationFace Recognition Based on SVM and 2DPCA
Vol. 4, o. 3, September, 2011 Face Recognton Based on SVM and 2DPCA Tha Hoang Le, Len Bu Faculty of Informaton Technology, HCMC Unversty of Scence Faculty of Informaton Scences and Engneerng, Unversty
More informationUser Authentication Based On Behavioral Mouse Dynamics Biometrics
User Authentcaton Based On Behavoral Mouse Dynamcs Bometrcs Chee-Hyung Yoon Danel Donghyun Km Department of Computer Scence Department of Computer Scence Stanford Unversty Stanford Unversty Stanford, CA
More informationSmoothing Spline ANOVA for variable screening
Smoothng Splne ANOVA for varable screenng a useful tool for metamodels tranng and mult-objectve optmzaton L. Rcco, E. Rgon, A. Turco Outlne RSM Introducton Possble couplng Test case MOO MOO wth Game Theory
More informationModular PCA Face Recognition Based on Weighted Average
odern Appled Scence odular PCA Face Recognton Based on Weghted Average Chengmao Han (Correspondng author) Department of athematcs, Lny Normal Unversty Lny 76005, Chna E-mal: hanchengmao@163.com Abstract
More informationMixed Linear System Estimation and Identification
48th IEEE Conference on Decson and Control, Shangha, Chna, December 2009 Mxed Lnear System Estmaton and Identfcaton A. Zymns S. Boyd D. Gornevsky Abstract We consder a mxed lnear system model, wth both
More informationControl strategies for network efficiency and resilience with route choice
Control strateges for networ effcency and reslence wth route choce Andy Chow Ru Sha Centre for Transport Studes Unversty College London, UK Centralsed strateges UK 1 Centralsed strateges Some effectve
More informationObject Recognition Based on Photometric Alignment Using Random Sample Consensus
Vol. 44 No. SIG 9(CVIM 7) July 2003 3 attached shadow photometrc algnment RANSAC RANdom SAmple Consensus Yale Face Database B RANSAC Object Recognton Based on Photometrc Algnment Usng Random Sample Consensus
More informationThe Research of Support Vector Machine in Agricultural Data Classification
The Research of Support Vector Machne n Agrcultural Data Classfcaton Le Sh, Qguo Duan, Xnmng Ma, Me Weng College of Informaton and Management Scence, HeNan Agrcultural Unversty, Zhengzhou 45000 Chna Zhengzhou
More informationMulti-stable Perception. Necker Cube
Mult-stable Percepton Necker Cube Spnnng dancer lluson, Nobuuk Kaahara Fttng and Algnment Computer Vson Szelsk 6.1 James Has Acknowledgment: Man sldes from Derek Hoem, Lana Lazebnk, and Grauman&Lebe 2008
More informationHuman Face Recognition Using Generalized. Kernel Fisher Discriminant
Human Face Recognton Usng Generalzed Kernel Fsher Dscrmnant ng-yu Sun,2 De-Shuang Huang Ln Guo. Insttute of Intellgent Machnes, Chnese Academy of Scences, P.O.ox 30, Hefe, Anhu, Chna. 2. Department of
More informationIntra-Parametric Analysis of a Fuzzy MOLP
Intra-Parametrc Analyss of a Fuzzy MOLP a MIAO-LING WANG a Department of Industral Engneerng and Management a Mnghsn Insttute of Technology and Hsnchu Tawan, ROC b HSIAO-FAN WANG b Insttute of Industral
More informationClassifier Selection Based on Data Complexity Measures *
Classfer Selecton Based on Data Complexty Measures * Edth Hernández-Reyes, J.A. Carrasco-Ochoa, and J.Fco. Martínez-Trndad Natonal Insttute for Astrophyscs, Optcs and Electroncs, Lus Enrque Erro No.1 Sta.
More informationLecture 4: Principal components
/3/6 Lecture 4: Prncpal components 3..6 Multvarate lnear regresson MLR s optmal for the estmaton data...but poor for handlng collnear data Covarance matrx s not nvertble (large condton number) Robustness
More informationWeighted Sparse Image Classification Based on Low Rank Representation
Copyrght 08 Tech Scence Press CMC, vol.56, no., pp.9-05, 08 Weghted Sparse Image Classfcaton Based on Low Rank Representaton Qd Wu, Ybng L, Yun Ln, * and Ruoln Zhou Abstract: The conventonal sparse representaton-based
More informationA Fast Visual Tracking Algorithm Based on Circle Pixels Matching
A Fast Vsual Trackng Algorthm Based on Crcle Pxels Matchng Zhqang Hou hou_zhq@sohu.com Chongzhao Han czhan@mal.xjtu.edu.cn Ln Zheng Abstract: A fast vsual trackng algorthm based on crcle pxels matchng
More informationSolving two-person zero-sum game by Matlab
Appled Mechancs and Materals Onlne: 2011-02-02 ISSN: 1662-7482, Vols. 50-51, pp 262-265 do:10.4028/www.scentfc.net/amm.50-51.262 2011 Trans Tech Publcatons, Swtzerland Solvng two-person zero-sum game by
More informationAn Optimal Algorithm for Prufer Codes *
J. Software Engneerng & Applcatons, 2009, 2: 111-115 do:10.4236/jsea.2009.22016 Publshed Onlne July 2009 (www.scrp.org/journal/jsea) An Optmal Algorthm for Prufer Codes * Xaodong Wang 1, 2, Le Wang 3,
More informationWavefront Reconstructor
A Dstrbuted Smplex B-Splne Based Wavefront Reconstructor Coen de Vsser and Mchel Verhaegen 14-12-201212 2012 Delft Unversty of Technology Contents Introducton Wavefront reconstructon usng Smplex B-Splnes
More informationCourse Introduction. Algorithm 8/31/2017. COSC 320 Advanced Data Structures and Algorithms. COSC 320 Advanced Data Structures and Algorithms
Course Introducton Course Topcs Exams, abs, Proects A quc loo at a few algorthms 1 Advanced Data Structures and Algorthms Descrpton: We are gong to dscuss algorthm complexty analyss, algorthm desgn technques
More informationImprovement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration
Improvement of Spatal Resoluton Usng BlockMatchng Based Moton Estmaton and Frame Integraton Danya Suga and Takayuk Hamamoto Graduate School of Engneerng, Tokyo Unversty of Scence, 6-3-1, Nuku, Katsuska-ku,
More informationA Unified Framework for Semantics and Feature Based Relevance Feedback in Image Retrieval Systems
A Unfed Framework for Semantcs and Feature Based Relevance Feedback n Image Retreval Systems Ye Lu *, Chunhu Hu 2, Xngquan Zhu 3*, HongJang Zhang 2, Qang Yang * School of Computng Scence Smon Fraser Unversty
More informationMachine Learning. K-means Algorithm
Macne Learnng CS 6375 --- Sprng 2015 Gaussan Mture Model GMM pectaton Mamzaton M Acknowledgement: some sldes adopted from Crstoper Bsop Vncent Ng. 1 K-means Algortm Specal case of M Goal: represent a data
More informationA CLASS OF TRANSFORMED EFFICIENT RATIO ESTIMATORS OF FINITE POPULATION MEAN. Department of Statistics, Islamia College, Peshawar, Pakistan 2
Pa. J. Statst. 5 Vol. 3(4), 353-36 A CLASS OF TRANSFORMED EFFICIENT RATIO ESTIMATORS OF FINITE POPULATION MEAN Sajjad Ahmad Khan, Hameed Al, Sadaf Manzoor and Alamgr Department of Statstcs, Islama College,
More informationMulti-View Face Alignment Using 3D Shape Model for View Estimation
Mult-Vew Face Algnment Usng 3D Shape Model for Vew Estmaton Yanchao Su 1, Hazhou A 1, Shhong Lao 1 Computer Scence and Technology Department, Tsnghua Unversty Core Technology Center, Omron Corporaton ahz@mal.tsnghua.edu.cn
More informationThe Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique
//00 :0 AM Outlne and Readng The Greedy Method The Greedy Method Technque (secton.) Fractonal Knapsack Problem (secton..) Task Schedulng (secton..) Mnmum Spannng Trees (secton.) Change Money Problem Greedy
More informationImage Alignment CSC 767
Image Algnment CSC 767 Image algnment Image from http://graphcs.cs.cmu.edu/courses/15-463/2010_fall/ Image algnment: Applcatons Panorama sttchng Image algnment: Applcatons Recognton of object nstances
More informationTN348: Openlab Module - Colocalization
TN348: Openlab Module - Colocalzaton Topc The Colocalzaton module provdes the faclty to vsualze and quantfy colocalzaton between pars of mages. The Colocalzaton wndow contans a prevew of the two mages
More informationContent Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers
IOSR Journal of Electroncs and Communcaton Engneerng (IOSR-JECE) e-issn: 78-834,p- ISSN: 78-8735.Volume 9, Issue, Ver. IV (Mar - Apr. 04), PP 0-07 Content Based Image Retreval Usng -D Dscrete Wavelet wth
More informationProblem Definitions and Evaluation Criteria for Computational Expensive Optimization
Problem efntons and Evaluaton Crtera for Computatonal Expensve Optmzaton B. Lu 1, Q. Chen and Q. Zhang 3, J. J. Lang 4, P. N. Suganthan, B. Y. Qu 6 1 epartment of Computng, Glyndwr Unversty, UK Faclty
More informationOutline. Type of Machine Learning. Examples of Application. Unsupervised Learning
Outlne Artfcal Intellgence and ts applcatons Lecture 8 Unsupervsed Learnng Professor Danel Yeung danyeung@eee.org Dr. Patrck Chan patrckchan@eee.org South Chna Unversty of Technology, Chna Introducton
More informationA Binarization Algorithm specialized on Document Images and Photos
A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a
More informationCategories and Subject Descriptors B.7.2 [Integrated Circuits]: Design Aids Verification. General Terms Algorithms
3. Fndng Determnstc Soluton from Underdetermned Equaton: Large-Scale Performance Modelng by Least Angle Regresson Xn L ECE Department, Carnege Mellon Unversty Forbs Avenue, Pttsburgh, PA 3 xnl@ece.cmu.edu
More informationAPPLICATION OF MULTIVARIATE LOSS FUNCTION FOR ASSESSMENT OF THE QUALITY OF TECHNOLOGICAL PROCESS MANAGEMENT
3. - 5. 5., Brno, Czech Republc, EU APPLICATION OF MULTIVARIATE LOSS FUNCTION FOR ASSESSMENT OF THE QUALITY OF TECHNOLOGICAL PROCESS MANAGEMENT Abstract Josef TOŠENOVSKÝ ) Lenka MONSPORTOVÁ ) Flp TOŠENOVSKÝ
More informationA Facet Generation Procedure. for solving 0/1 integer programs
A Facet Generaton Procedure for solvng 0/ nteger programs by Gyana R. Parja IBM Corporaton, Poughkeepse, NY 260 Radu Gaddov Emery Worldwde Arlnes, Vandala, Oho 45377 and Wlbert E. Wlhelm Teas A&M Unversty,
More informationCLASSIFICATION OF ULTRASONIC SIGNALS
The 8 th Internatonal Conference of the Slovenan Socety for Non-Destructve Testng»Applcaton of Contemporary Non-Destructve Testng n Engneerng«September -3, 5, Portorož, Slovena, pp. 7-33 CLASSIFICATION
More informationCompiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz
Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster
More informationy and the total sum of
Lnear regresson Testng for non-lnearty In analytcal chemstry, lnear regresson s commonly used n the constructon of calbraton functons requred for analytcal technques such as gas chromatography, atomc absorpton
More informationCS 534: Computer Vision Model Fitting
CS 534: Computer Vson Model Fttng Sprng 004 Ahmed Elgammal Dept of Computer Scence CS 534 Model Fttng - 1 Outlnes Model fttng s mportant Least-squares fttng Maxmum lkelhood estmaton MAP estmaton Robust
More informationEnhanced AMBTC for Image Compression using Block Classification and Interpolation
Internatonal Journal of Computer Applcatons (0975 8887) Volume 5 No.0, August 0 Enhanced AMBTC for Image Compresson usng Block Classfcaton and Interpolaton S. Vmala Dept. of Comp. Scence Mother Teresa
More informationStereo Depth Continuity
Stereo Depth Contnuty Steven Damond (stevend@stanford.edu), Jessca Taylor (jacobt@stanford.edu) March 17, 014 1 Abstract We tackle the problem of producng depth maps from stereo vdeo. Some algorthms for
More informationAngle-Independent 3D Reconstruction. Ji Zhang Mireille Boutin Daniel Aliaga
Angle-Independent 3D Reconstructon J Zhang Mrelle Boutn Danel Alaga Goal: Structure from Moton To reconstruct the 3D geometry of a scene from a set of pctures (e.g. a move of the scene pont reconstructon
More informationNovel Fuzzy logic Based Edge Detection Technique
Novel Fuzzy logc Based Edge Detecton Technque Aborsade, D.O Department of Electroncs Engneerng, adoke Akntola Unversty of Tech., Ogbomoso. Oyo-state. doaborsade@yahoo.com Abstract Ths paper s based on
More informationEfficient Distributed Linear Classification Algorithms via the Alternating Direction Method of Multipliers
Effcent Dstrbuted Lnear Classfcaton Algorthms va the Alternatng Drecton Method of Multplers Caoxe Zhang Honglak Lee Kang G. Shn Department of EECS Unversty of Mchgan Ann Arbor, MI 48109, USA caoxezh@umch.edu
More informationClassifying Acoustic Transient Signals Using Artificial Intelligence
Classfyng Acoustc Transent Sgnals Usng Artfcal Intellgence Steve Sutton, Unversty of North Carolna At Wlmngton (suttons@charter.net) Greg Huff, Unversty of North Carolna At Wlmngton (jgh7476@uncwl.edu)
More informationSVM-based Learning for Multiple Model Estimation
SVM-based Learnng for Multple Model Estmaton Vladmr Cherkassky and Yunqan Ma Department of Electrcal and Computer Engneerng Unversty of Mnnesota Mnneapols, MN 55455 {cherkass,myq}@ece.umn.edu Abstract:
More informationCS246: Mining Massive Datasets Jure Leskovec, Stanford University
CS46: Mnng Massve Datasets Jure Leskovec, Stanford Unversty http://cs46.stanford.edu /19/013 Jure Leskovec, Stanford CS46: Mnng Massve Datasets, http://cs46.stanford.edu Perceptron: y = sgn( x Ho to fnd
More informationMachine Learning. Topic 6: Clustering
Machne Learnng Topc 6: lusterng lusterng Groupng data nto (hopefully useful) sets. Thngs on the left Thngs on the rght Applcatons of lusterng Hypothess Generaton lusters mght suggest natural groups. Hypothess
More informationSuper-resolution with Nonlocal Regularized Sparse Representation
Super-resoluton wth Nonlocal Regularzed Sparse Representaton Wesheng Dong a, Guangmng Sh a, Le Zhang b, and Xaoln Wu c a Key Laboratory of Intellgent Percepton and Image Understandng (Chnese Mnstry of
More informationParallel Numerics. 1 Preconditioning & Iterative Solvers (From 2016)
Technsche Unverstät München WSe 6/7 Insttut für Informatk Prof. Dr. Thomas Huckle Dpl.-Math. Benjamn Uekermann Parallel Numercs Exercse : Prevous Exam Questons Precondtonng & Iteratve Solvers (From 6)
More informationA Probabilistic Approach to Detect Urban Regions from Remotely Sensed Images Based on Combination of Local Features
A Probablstc Approach to Detect Urban Regons from Remotely Sensed Images Based on Combnaton of Local Features Berl Sırmaçek German Aerospace Center (DLR) Remote Sensng Technology Insttute Weßlng, 82234,
More informationSum of Linear and Fractional Multiobjective Programming Problem under Fuzzy Rules Constraints
Australan Journal of Basc and Appled Scences, 2(4): 1204-1208, 2008 ISSN 1991-8178 Sum of Lnear and Fractonal Multobjectve Programmng Problem under Fuzzy Rules Constrants 1 2 Sanjay Jan and Kalash Lachhwan
More informationFace Recognition Method Based on Within-class Clustering SVM
Face Recognton Method Based on Wthn-class Clusterng SVM Yan Wu, Xao Yao and Yng Xa Department of Computer Scence and Engneerng Tong Unversty Shangha, Chna Abstract - A face recognton method based on Wthn-class
More informationOptimizing Document Scoring for Query Retrieval
Optmzng Document Scorng for Query Retreval Brent Ellwen baellwe@cs.stanford.edu Abstract The goal of ths project was to automate the process of tunng a document query engne. Specfcally, I used machne learnng
More informationUnsupervised Learning and Clustering
Unsupervsed Learnng and Clusterng Supervsed vs. Unsupervsed Learnng Up to now we consdered supervsed learnng scenaro, where we are gven 1. samples 1,, n 2. class labels for all samples 1,, n Ths s also
More informationHermite Splines in Lie Groups as Products of Geodesics
Hermte Splnes n Le Groups as Products of Geodescs Ethan Eade Updated May 28, 2017 1 Introducton 1.1 Goal Ths document defnes a curve n the Le group G parametrzed by tme and by structural parameters n the
More informationNetwork Intrusion Detection Based on PSO-SVM
TELKOMNIKA Indonesan Journal of Electrcal Engneerng Vol.1, No., February 014, pp. 150 ~ 1508 DOI: http://dx.do.org/10.11591/telkomnka.v1.386 150 Network Intruson Detecton Based on PSO-SVM Changsheng Xang*
More informationLocal Quaternary Patterns and Feature Local Quaternary Patterns
Local Quaternary Patterns and Feature Local Quaternary Patterns Jayu Gu and Chengjun Lu The Department of Computer Scence, New Jersey Insttute of Technology, Newark, NJ 0102, USA Abstract - Ths paper presents
More informationSubspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;
Subspace clusterng Clusterng Fundamental to all clusterng technques s the choce of dstance measure between data ponts; D q ( ) ( ) 2 x x = x x, j k = 1 k jk Squared Eucldean dstance Assumpton: All features
More informationFEATURE EXTRACTION. Dr. K.Vijayarekha. Associate Dean School of Electrical and Electronics Engineering SASTRA University, Thanjavur
FEATURE EXTRACTION Dr. K.Vjayarekha Assocate Dean School of Electrcal and Electroncs Engneerng SASTRA Unversty, Thanjavur613 41 Jont Intatve of IITs and IISc Funded by MHRD Page 1 of 8 Table of Contents
More informationFast Feature Value Searching for Face Detection
Vol., No. 2 Computer and Informaton Scence Fast Feature Value Searchng for Face Detecton Yunyang Yan Department of Computer Engneerng Huayn Insttute of Technology Hua an 22300, Chna E-mal: areyyyke@63.com
More informationImage Deblurring and Super-resolution by Adaptive Sparse Domain Selection and Adaptive Regularization
Image Deblurrng and Super-resoluton by Adaptve Sparse Doman Selecton and Adaptve Regularzaton Wesheng Dong a,b, Le Zhang b,1, Member, IEEE, Guangmng Sh a, Senor Member, IEEE, and Xaoln Wu c, Senor Member,
More information6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour
6.854 Advanced Algorthms Petar Maymounkov Problem Set 11 (November 23, 2005) Wth: Benjamn Rossman, Oren Wemann, and Pouya Kheradpour Problem 1. We reduce vertex cover to MAX-SAT wth weghts, such that the
More informationUnsupervised Learning
Pattern Recognton Lecture 8 Outlne Introducton Unsupervsed Learnng Parametrc VS Non-Parametrc Approach Mxture of Denstes Maxmum-Lkelhood Estmates Clusterng Prof. Danel Yeung School of Computer Scence and
More informationMULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION
MULTISPECTRAL IMAGES CLASSIFICATION BASED ON KLT AND ATR AUTOMATIC TARGET RECOGNITION Paulo Quntlano 1 & Antono Santa-Rosa 1 Federal Polce Department, Brasla, Brazl. E-mals: quntlano.pqs@dpf.gov.br and
More informationCorner-Based Image Alignment using Pyramid Structure with Gradient Vector Similarity
Journal of Sgnal and Informaton Processng, 013, 4, 114-119 do:10.436/jsp.013.43b00 Publshed Onlne August 013 (http://www.scrp.org/journal/jsp) Corner-Based Image Algnment usng Pyramd Structure wth Gradent
More information