Discriminative Improvements to Distributional Sentence Similarity

Size: px
Start display at page:

Download "Discriminative Improvements to Distributional Sentence Similarity"

Transcription

1 Dscrmnatve Improvements to Dstrbutonal Sentence Smlarty Yangfeng J School of Interactve Computng Georga Insttute of Technology [email protected] Jacob Esensten School of Interactve Computng Georga Insttute of Technology [email protected] Abstract Matrx and tensor factorzaton have been appled to a number of semantc relatedness tasks, ncludng paraphrase dentfcaton. The key dea s that smlarty n the latent space mples semantc relatedness. We descrbe three ways n whch labeled data can mprove the accuracy of these approaches on paraphrase classfcaton. Frst, we desgn a new dscrmnatve term-weghtng metrc called TF-KLD, whch outperforms TF-IDF. Next, we show that usng the latent representaton from matrx factorzaton as features n a classfcaton algorthm substantally mproves accuracy. Fnally, we combne latent features wth fne-graned n-gram overlap features, yeldng performance that s 3% more accurate than the pror state-of-the-art. 1 Introducton Measurng the semantc smlarty of short unts of text s fundamental to many natural language processng tasks, from evaluatng machne translaton (Kauchak and Barzlay, 2006) to groupng redundant event mentons n socal meda (Petrovć et al., 2010). The task s challengng because of the nfntely dverse set of possble lngustc realzatons for any dea (Bhagat and Hovy, 2013), and because of the short length of ndvdual sentences, whch means that standard bag-of-words representatons wll be hopelessly sparse. Dstrbutonal methods address ths problem by transformng the hgh-dmensonal bag-of-words representaton nto a lower-dmensonal latent space. Ths can be accomplshed by factorng a matrx or tensor of term-context counts (Turney and Pantel, 2010); proxmty n the nduced latent space has been shown to correlate wth semantc smlarty (Mhalcea et al., 2006). However, factorng the term-context matrx means throwng away a consderable amount of nformaton, as the orgnal matrx of sze M N (number of nstances by number of features) s factored nto two smaller matrces of sze M K and N K, wth K M, N. If the factorzaton does not take nto account labeled data about semantc smlarty, mportant nformaton can be lost. In ths paper, we show how labeled data can consderably mprove dstrbutonal methods for measurng semantc smlarty. Frst, we develop a new dscrmnatve term-weghtng metrc called TF-KLD, whch s appled to the term-context matrx before factorzaton. On a standard paraphrase dentfcaton task (Dolan et al., 2004), ths method mproves on both tradtonal TF-IDF and Weghted Textual Matrx Factorzaton (WTMF; Guo and Dab, 2012). Next, we convert the latent representatons of each sentence par nto a feature vector, whch s used as nput to a lnear SVM classfer. Ths yelds further mprovements and substantally outperforms the current state-of-the-art on paraphrase classfcaton. We then add fnegraned features about the lexcal smlarty of the sentence par. The combnaton of latent and fnegraned features yelds further mprovements n accuracy, demonstratng that these feature sets provde complementary nformaton on semantc smlarty.

2 2 Related Work Wthout attemptng to do justce to the entre lterature on paraphrase dentfcaton, we note three hgh-level approaches: (1) strng smlarty metrcs such as n-gram overlap and BLEU score (Wan et al., 2006; Madnan et al., 2012), as well as strng kernels (Bu et al., 2012); (2) syntactc operatons on the parse structure (Wu, 2005; Das and Smth, 2009); and (3) dstrbutonal methods, such as latent semantc analyss (LSA; Landauer et al., 1998), whch are most relevant to our work. One applcaton of dstrbutonal technques s to replace ndvdual words wth dstrbutonally smlar alternatves (Kauchak and Barzlay, 2006). Alternatvely, Blacoe and Lapata (2012) show that latent word representatons can be combned wth smple elementwse operatons to dentfy the semantc smlarty of larger unts of text. Socher et al. (2011) propose a syntactcally-nformed approach to combne word representatons, usng a recursve auto-encoder to propagate meanng through the parse tree. We take a dfferent approach: rather than representng the meanngs of ndvdual words, we drectly obtan a dstrbutonal representaton for the entre sentence. Ths s nspred by Mhalcea et al. (2006) and Guo and Dab (2012), who treat sentences as pseudo-documents n an LSA framework, and dentfy paraphrases usng smlarty n the latent space. We show that the performance of such technques can be mproved dramatcally by usng supervsed nformaton to (1) reweght the ndvdual dstrbutonal features and (2) learn the mportance of each latent dmenson. 3 Dscrmnatve feature weghtng Dstrbutonal representatons (Turney and Pantel, 2010) can be nduced from a co-occurrence matrx W R M N, where M s the number of nstances and N s the number of dstrbutonal features. For paraphrase dentfcaton, each nstance s a sentence; features may be ungrams, or may nclude hgher-order n-grams or dependency pars. By decomposng the matrx W, we hope to obtan a latent representaton n whch semantcally-related sentences are smlar. Sngular value decomposton (SVD) s tradtonally used to perform ths factorzaton. However, recent work has demonstrated the robustness of nonnegatve matrx factorzaton (NMF; Lee and Seung, 2001) for text mnng tasks (Xu et al., 2003; Arora et al., 2012); the dfference from SVD s the addton of a non-negatvty constrant n the latent representaton based on non-orthogonal bass. Whle W may smply contan counts of dstrbutonal features, pror work has demonstrated the utlty of reweghtng these counts (Turney and Pantel, 2010). TF-IDF s a standard approach, as the nverse document frequency (IDF) term ncreases the mportance of rare words, whch may be more dscrmnatve. Guo and Dab (2012) show that applyng a specal weght to unseen words can further mprovement performance on paraphrase dentfcaton. We present a new weghtng scheme, TF-KLD, based on supervsed nformaton. The key dea s to ncrease the weghts of dstrbutonal features that are dscrmnatve, and to decrease the weghts of features that are not. Conceptually, ths s smlar to Lnear Dscrmnant Analyss, a supervsed feature weghtng scheme for contnuous data (Murphy, 2012). More formally, we assume labeled sentence pars of the form w (1), w (2), r, where w (1) s the vector of dstrbutonal features for the frst sentence, w (2) s the vector of dstrbutonal features for the second sentence, and r {0, 1} ndcates whether they are labeled as a paraphrase par. Assumng the order of the sentences wthn the par s rrelevant, then for k-th dstrbutonal feature, we defne two Bernoull dstrbutons: p k = P (w (1) k w(2) k = 1, r = 1). Ths s the probablty that sentence w (1) contans feature k, gven that k appears n w (2) and the two sentences are labeled as paraphrases, r = 1. q k = P (w (1) k w(2) k = 1, r = 0). Ths s the probablty that sentence w (1) contans feature k, gven that k appears n w (2) and the two sentences are labeled as not paraphrases, r = 0. The Kullback-Lebler dvergence KL(p k q k ) = x p k(x) log p k(x) q k (x) s then a measure of the dscrmnablty of feature k, and s guaranteed to be non-

3 q k off then study same nether shares not but fear nor ungram recall 2 ungram precson 3 bgram recall 4 bgram precson 5 dependency relaton recall 6 dependency relaton precson 7 BLEU recall 8 BLEU precson 9 Dfference of sentence length 10 Tree-edtng dstance p k Fgure 1: Condtonal probabltes for a few handselected ungram features, wth lnes showng contours wth dentcal KL-dvergence. The probabltes are estmated based on the MSRPC tranng set (Dolan et al., 2004). negatve. 1 We use ths dvergence to reweght the features n W before performng the matrx factorzaton. Ths has the effect of ncreasng the weghts of features whose lkelhood of appearng n a par of sentences s strongly nfluenced by the paraphrase relatonshp between the two sentences. On the other hand, f p k = q k, then the KL-dvergence wll be zero, and the feature wll be gnored n the matrx factorzaton. We name ths weghtng scheme TF-KLD, snce t ncludes the term frequency and the KL-dvergence. Takng the ungram feature not as an example, we have p k = [0.66, 0.34] and q k = [0.31, 0.69], for a KL-dvergence of 0.25: the lkelhood of ths word beng shared between two sentence s strongly dependent on whether the sentences are paraphrases. In contrast, the feature then has p k = [0.33, 0.67] and q k = [0.32, 0.68], for a KL-dvergence of Fgure 1 shows the dstrbutons of these and other ungram features wth respect to p k and 1 q k. The dagonal lne runnng through the mddle of the plot ndcates zero KL-dvergence, so features on ths lne wll be gnored. 1 We obtan very smlar results wth the opposte dvergence KL(q k p k ). However, the symmetrc Jensen-Shannon dvergence performs poorly. Table 1: Fne-graned features for paraphrase classfcaton, selected from pror work (Wan et al., 2006). 4 Supervsed classfcaton Whle prevous work has performed paraphrase classfcaton usng dstance or smlarty n the latent space (Guo and Dab, 2012; Socher et al., 2011), more drect supervson can be appled. Specfcally, we convert the latent representatons of a par of sentences v 1 and v 2 nto a sample vector, s( v 1, v 2 ) = [ v 1 + v 2, v 1 v 2 ], (1) concatenatng the element-wse sum v 1 + v 2 and absolute dfference v 1 v 2. Note that s(, ) s symmetrc, snce s( v 1, v 2 ) = s( v 2, v 1 ). Gven ths representaton, we can use any supervsed classfcaton algorthm. A further advantage of treatng paraphrase as a supervsed classfcaton problem s that we can apply addtonal features besdes the latent representaton. We consder a subset of features dentfed by Wan et al. (2006), lsted n Table 1. These features manly capture fne-graned smlarty between sentences, for example by countng specfc ungram and bgram overlap. 5 Experments Our experments test the utlty of the TF- KLD weghtng towards paraphrase classfcaton, usng the Mcrosoft Research Paraphrase Corpus (Dolan et al., 2004). The tranng set contans 2753 true paraphrase pars and 1323 false paraphrase pars; the test set contans 1147 and 578 pars, respectvely. The TF-KLD weghts are constructed from only the tranng set, whle matrx factorzatons are per-

4 formed on the entre corpus. Matrx factorzaton on both tranng and (unlabeled) test data can be vewed as a form of transductve learnng (Gammerman et al., 1998), where we assume access to unlabeled test set nstances. 2 We also consder an nductve settng, where we construct the bass of the latent space from only the tranng set, and then project the test set onto ths bass to fnd the correspondng latent representaton. The performance dfferences between the transductve and nductve settngs were generally between 0.5% and 1%, as noted n detal below. We reterate that the TF-KLD weghts are never computed from test set data. Pror work on ths dataset s descrbed n secton 2. To our knowledge, the current state-of-theart s a supervsed system that combnes several machne translaton metrcs (Madnan et al., 2012), but we also compare wth state-of-the-art unsupervsed matrx factorzaton work (Guo and Dab, 2012). 5.1 Smlarty-based classfcaton In the frst experment, we predct whether a par of sentences s a paraphrase by measurng ther cosne smlarty n latent space, usng a threshold for the classfcaton boundary. As n pror work (Guo and Dab, 2012), the threshold s tuned on held-out tranng data. We consder two dstrbutonal feature sets: FEAT 1, whch ncludes ungrams; and FEAT 2, whch also ncludes bgrams and unlabeled dependency pars obtaned from MaltParser (Nvre et al., 2007). To compare wth Guo and Dab (2012), we set the latent dmensonalty to K = 100, whch was the same n ther paper. Both SVD and NMF factorzaton are evaluated; n both cases, we mnmze the Frobenus norm of the reconstructon error. Table 2 compares the accuracy of a number of dfferent confguratons. The transductve TF-KLD weghtng yelds the best overall accuracy, achevng 72.75% when combned wth nonnegatve matrx factorzaton. Whle NMF performs slghtly better than SVD n both comparsons, the major dfference s the performance of dscrmnatve TF-KLD weghtng, whch outperforms TF-IDF regardless of the factorzaton technque. When we 2 Another example of transductve learnng n NLP s when Turan et al. (2010) nduced word representatons from a corpus that ncluded both tranng and test data for ther downstream named entty recognton task. Accuracy (%) Feat 1 _TF-IDF_SVM Feat 2 _TF-IDF_SVM Feat 1 _TF-KLD_SVM Feat 2 _TF-KLD_SVM K Fgure 2: Accuracy of feature and weghtng combnatons n the classfcaton framework. perform the matrx factorzaton on only the tranng data, the accuracy on the test set s 73.58%, wth F1 score 80.55%. 5.2 Supervsed classfcaton Next, we apply supervsed classfcaton, constructng sample vectors from the latent representaton as shown n Equaton 1. For classfcaton, we choose a Support Vector Machne wth a lnear kernel (Fan et al., 2008), leavng a thorough comparson of classfers for future work. The classfer parameter C s tuned on a development set comprsng 20% of the orgnal tranng set. Fgure 2 presents results for a range of latent dmensonaltes. Supervsed learnng dentfes the mportant dmensons n the latent space, yeldng sgnfcantly better performance that the smlartybased classfcaton from the prevous experment. In Table 3, we compare aganst pror publshed work, usng the held-out development set to select the best value of K (agan, K = 400). The best result s from TF-KLD, wth dstrbutonal features FEAT 2, achevng 79.76% accuracy and 85.87% F1. Ths s well beyond all known pror results on ths task. When we nduce the latent bass from only the tranng data, we get 78.55% on accuracy and 84.59% F1, also better than the prevous state-of-art. Fnally, we augment the dstrbutonal representaton, concatenatng the ten fne-graned features n Table 1 to the sample vectors descrbed n Equaton 1. As shown n Table 3, the accu-

5 Factorzaton Feature set Weghtng K Measure Accuracy (%) F1 SVD ungrams TF-IDF 100 cosne sm NMF ungrams TF-IDF 100 cosne sm WTMF ungrams TF-IDF 100 cosne sm not reported SVD ungrams TF-KLD 100 cosne sm NMF ungrams TF-KLD 100 cosne sm Table 2: Smlarty-based paraphrase dentfcaton accuracy. Results for WTMF are reprnted from the paper by Guo and Dab (2012). Acc. F1 Most common class (Wan et al., 2006) (Das and Smth, 2009) (Das and Smth, 2009) wth 18 features (Bu et al., 2012) 76.3 not reported (Socher et al., 2011) (Madnan et al., 2012) FEAT 2, TF-KLD, SVM FEAT 2, TF-KLD, SVM, Fne-graned features Table 3: Supervsed classfcaton. Results from pror work are reprnted. racy now mproves to 80.41%, wth an F1 score of 85.96%. When the latent representaton s nduced from only the tranng data, the correspondng results are 79.94% on accuracy and 85.36% F1, agan better than the prevous state-of-the-art. These results show that the nformaton captured by the dstrbutonal representaton can stll be augmented by more fne-graned tradtonal features. 6 Concluson We have presented three ways n whch labeled data can mprove dstrbutonal measures of semantc smlarty at the sentence level. The man nnovaton s TF-KLD, whch dscrmnatvely reweghts the dstrbutonal features before factorzaton, so that dscrmnablty mpacts the nducton of the latent representaton. We then transform the latent representaton nto a sample vector for supervsed learnng, obtanng results that strongly outperform the pror state-of-the-art; addng fne-graned lexcal features further ncreases performance. These deas may have applcablty n other semantc smlarty tasks, and we are also eager to apply them to new, large-scale automatcally-nduced paraphrase corpora (Gantkevtch et al., 2013). Acknowledgments We thank the revewers for ther helpful feedback, and Wewe Guo for quckly answerng questons about hs mplementaton. Ths research was supported by a Google Faculty Research Award to the second author. References Sanjeev Arora, Rong Ge, and Ankur Motra Learnng Topc Models - Gong beyond SVD. In FOCS, pages Rahul Bhagat and Eduard Hovy What Is a Paraphrase? Computatonal Lngustcs. Wllam Blacoe and Mrella Lapata A Comparson of Vector-based Representatons for Semantc Composton. In Proceedngs of the 2012 Jont Conference on Emprcal Methods n Natural Language Processng and Computatonal Natural Language Learnng, pages , Stroudsburg, PA, USA. Assocaton for Computatonal Lngustcs. Fan Bu, Hang L, and Xaoyan Zhu Strng Rewrtng kernel. In Proceedngs of ACL, pages Assocaton for Computatonal Lngustcs. Dpanjan Das and Noah A. Smth Paraphrase dentfcaton as probablstc quas-synchronous recognton. In Proceedngs of the Jont Conference

6 of the Annual Meetng of the Assocaton for Computatonal Lngustcs and the Internatonal Jont Conference on Natural Language Processng, pages , Stroudsburg, PA, USA. Assocaton for Computatonal Lngustcs. Bll Dolan, Chrs Qurk, and Chrs Brockett Unsupervsed Constructon of Large Paraphrase Corpora: Explotng Massvely Parallel News Sources. In COL- ING. Rong-En Fan, Ka-We Chang, Cho-Ju Hseh, Xang-Ru Wang, and Chh-Jen Ln LIBLINEAR: A Lbrary for Large Lnear Classfcaton. Journal of Machne Learnng Research, 9: Alexander Gammerman, Volodya Vovk, and Vladmr Vapnk Learnng by transducton. In Proceedngs of the Fourteenth conference on Uncertanty n artfcal ntellgence, pages Morgan Kaufmann Publshers Inc. Jur Gantkevtch, Benjamn Van Durme, and Chrs Callson-Burch PPDB: The Paraphrase Database. In Proceedngs of NAACL, pages Assocaton for Computatonal Lngustcs. Wewe Guo and Mona Dab Modelng Sentences n the Latent Space. In Proceedngs of the 50th Annual Meetng of the Assocaton for Computatonal Lngustcs, pages , Stroudsburg, PA, USA. Assocaton for Computatonal Lngustcs. Davd Kauchak and Regna Barzlay Paraphrasng for automatc evaluaton. In Proceedngs of NAACL, pages Assocaton for Computatonal Lngustcs. Thomas Landauer, Peter W. Foltz, and Darrel Laham Introducton to Latent Semantc Analyss. Dscource Processes, 25: Danel D. Lee and H. Sebastan Seung Algorthms for Non-Negatve Matrx Factorzaton. In Advances n Neural Informaton Processng Systems (NIPS). Ntn Madnan, Joel R. Tetreault, and Martn Chodorow Re-examnng Machne Translaton Metrcs for Paraphrase Identfcaton. In HLT-NAACL, pages The Assocaton for Computatonal Lngustcs. Rada Mhalcea, Courtney Corley, and Carlo Strapparava Corpus-based and knowledge-based measures of text semantc smlarty. In AAAI. Kevn P. Murphy Machne Learnng: A Probablstc Perspectve. The MIT Press. Joakm Nvre, Johan Hall, Jens Nlsson, Atanas Chanev, Gülsen Erygt, Sandra Kübler, Svetoslav Marnov, and Erwn Mars MaltParser: A languagendependent system for data-drven dependency parsng. Natural Language Engneerng, 13(2): Saša Petrovć, Mles Osborne, and Vctor Lavrenko Streamng frst story detecton wth applcaton to twtter. In Proceedngs of HLT-NAACL, pages Assocaton for Computatonal Lngustcs. Rchard Socher, Erc H. Huang, Jeffrey Pennngton, Andrew Y. Ng, and Chrstopher D. Mannng Dynamc Poolng And Unfoldng Recursve Autoencoders For Paraphrase Detecton. In Advances n Neural Informaton Processng Systems (NIPS). Joseph Turan, Lev Ratnov, and Yoshua Bengo Word Representaton: A Smple and General Method for Sem-Supervsed Learnng. In ACL, pages Peter D. Turney and Patrck Pantel From Frequency to Meanng: Vector Space Models of Semantcs. JAIR, 37: Ssephen Wan, Mark Dras, Robert Dale, and Cecle Pars Usng Dependency-based Features to Take the Para-farce out of Paraphrase. In Proceedngs of the Australasan Language Technology Workshop. Deka Wu Recognzng paraphrases and textual entalment usng nverson transducton grammars. In Proceedngs of the ACL Workshop on Emprcal Modelng of Semantc Equvalence and Entalment, pages Assocaton for Computatonal Lngustcs. We Xu, Xn Lu, and Yhong Gong Document Clusterng based on Non-Negatve Matrx Factorzaton. In SIGIR, pages

Face Verification Problem. Face Recognition Problem. Application: Access Control. Biometric Authentication. Face Verification (1:1 matching)

Face Verification Problem. Face Recognition Problem. Application: Access Control. Biometric Authentication. Face Verification (1:1 matching) Face Recognton Problem Face Verfcaton Problem Face Verfcaton (1:1 matchng) Querymage face query Face Recognton (1:N matchng) database Applcaton: Access Control www.vsage.com www.vsoncs.com Bometrc Authentcaton

More information

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis The Development of Web Log Mnng Based on Improve-K-Means Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna [email protected] Abstract.

More information

Forecasting the Direction and Strength of Stock Market Movement

Forecasting the Direction and Strength of Stock Market Movement Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye [email protected] [email protected] [email protected] Abstract - Stock market s one of the most complcated systems

More information

Learning from Multiple Outlooks

Learning from Multiple Outlooks Learnng from Multple Outlooks Maayan Harel Department of Electrcal Engneerng, Technon, Hafa, Israel She Mannor Department of Electrcal Engneerng, Technon, Hafa, Israel [email protected] [email protected]

More information

DEFINING %COMPLETE IN MICROSOFT PROJECT

DEFINING %COMPLETE IN MICROSOFT PROJECT CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMI-SP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,

More information

What is Candidate Sampling

What is Candidate Sampling What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble

More information

Conversion between the vector and raster data structures using Fuzzy Geographical Entities

Conversion between the vector and raster data structures using Fuzzy Geographical Entities Converson between the vector and raster data structures usng Fuzzy Geographcal Enttes Cdála Fonte Department of Mathematcs Faculty of Scences and Technology Unversty of Combra, Apartado 38, 3 454 Combra,

More information

Comparison of Domain-Specific Lexicon Construction Methods for Sentiment Analysis

Comparison of Domain-Specific Lexicon Construction Methods for Sentiment Analysis , pp.152-156 http://d.do.org/10.14257/astl.2016.135.38 Comparson of Doman-Specfc Lecon Constructon Methods for Sentment Analyss Myeong So Km 1, Jong Woo Km 2,3 and Cu Jng 4 1 Department of Mathematcs,

More information

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network 700 Proceedngs of the 8th Internatonal Conference on Innovaton & Management Forecastng the Demand of Emergency Supples: Based on the CBR Theory and BP Neural Network Fu Deqang, Lu Yun, L Changbng School

More information

Web Object Indexing Using Domain Knowledge *

Web Object Indexing Using Domain Knowledge * Web Object Indexng Usng Doman Knowledge * Muyuan Wang Department of Automaton Tsnghua Unversty Bejng 100084, Chna (86-10)51774518 Zhwe L, Le Lu, We-Yng Ma Mcrosoft Research Asa Sgma Center, Hadan Dstrct

More information

An Interest-Oriented Network Evolution Mechanism for Online Communities

An Interest-Oriented Network Evolution Mechanism for Online Communities An Interest-Orented Network Evoluton Mechansm for Onlne Communtes Cahong Sun and Xaopng Yang School of Informaton, Renmn Unversty of Chna, Bejng 100872, P.R. Chna {chsun,yang}@ruc.edu.cn Abstract. Onlne

More information

Assessing Student Learning Through Keyword Density Analysis of Online Class Messages

Assessing Student Learning Through Keyword Density Analysis of Online Class Messages Assessng Student Learnng Through Keyword Densty Analyss of Onlne Class Messages Xn Chen New Jersey Insttute of Technology [email protected] Brook Wu New Jersey Insttute of Technology [email protected] ABSTRACT Ths

More information

Out-of-Sample Extensions for LLE, Isomap, MDS, Eigenmaps, and Spectral Clustering

Out-of-Sample Extensions for LLE, Isomap, MDS, Eigenmaps, and Spectral Clustering Out-of-Sample Extensons for LLE, Isomap, MDS, Egenmaps, and Spectral Clusterng Yoshua Bengo, Jean-Franços Paement, Pascal Vncent Olver Delalleau, Ncolas Le Roux and Mare Oumet Département d Informatque

More information

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by 6 CHAPTER 8 COMPLEX VECTOR SPACES 5. Fnd the kernel of the lnear transformaton gven n Exercse 5. In Exercses 55 and 56, fnd the mage of v, for the ndcated composton, where and are gven by the followng

More information

L10: Linear discriminants analysis

L10: Linear discriminants analysis L0: Lnear dscrmnants analyss Lnear dscrmnant analyss, two classes Lnear dscrmnant analyss, C classes LDA vs. PCA Lmtatons of LDA Varants of LDA Other dmensonalty reducton methods CSCE 666 Pattern Analyss

More information

Vision Mouse. Saurabh Sarkar a* University of Cincinnati, Cincinnati, USA ABSTRACT 1. INTRODUCTION

Vision Mouse. Saurabh Sarkar a* University of Cincinnati, Cincinnati, USA ABSTRACT 1. INTRODUCTION Vson Mouse Saurabh Sarkar a* a Unversty of Cncnnat, Cncnnat, USA ABSTRACT The report dscusses a vson based approach towards trackng of eyes and fngers. The report descrbes the process of locatng the possble

More information

GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM

GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM GRAVITY DATA VALIDATION AND OUTLIER DETECTION USING L 1 -NORM BARRIOT Jean-Perre, SARRAILH Mchel BGI/CNES 18.av.E.Beln 31401 TOULOUSE Cedex 4 (France) Emal: [email protected] 1/Introducton The

More information

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic Lagrange Multplers as Quanttatve Indcators n Economcs Ivan Mezník Insttute of Informatcs, Faculty of Busness and Management, Brno Unversty of TechnologCzech Republc Abstract The quanttatve role of Lagrange

More information

Logistic Regression. Steve Kroon

Logistic Regression. Steve Kroon Logstc Regresson Steve Kroon Course notes sectons: 24.3-24.4 Dsclamer: these notes do not explctly ndcate whether values are vectors or scalars, but expects the reader to dscern ths from the context. Scenaro

More information

Single and multiple stage classifiers implementing logistic discrimination

Single and multiple stage classifiers implementing logistic discrimination Sngle and multple stage classfers mplementng logstc dscrmnaton Hélo Radke Bttencourt 1 Dens Alter de Olvera Moraes 2 Vctor Haertel 2 1 Pontfíca Unversdade Católca do Ro Grande do Sul - PUCRS Av. Ipranga,

More information

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..

More information

v a 1 b 1 i, a 2 b 2 i,..., a n b n i.

v a 1 b 1 i, a 2 b 2 i,..., a n b n i. SECTION 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS 455 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS All the vector spaces we have studed thus far n the text are real vector spaces snce the scalars are

More information

Support Vector Machines

Support Vector Machines Support Vector Machnes Max Wellng Department of Computer Scence Unversty of Toronto 10 Kng s College Road Toronto, M5S 3G5 Canada [email protected] Abstract Ths s a note to explan support vector machnes.

More information

Semantic Link Analysis for Finding Answer Experts *

Semantic Link Analysis for Finding Answer Experts * JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 28, 51-65 (2012) Semantc Lnk Analyss for Fndng Answer Experts * YAO LU 1,2,3, XIAOJUN QUAN 2, JINGSHENG LEI 4, XINGLIANG NI 1,2,3, WENYIN LIU 2,3 AND YINLONG

More information

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ). REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or

More information

Performance Analysis and Coding Strategy of ECOC SVMs

Performance Analysis and Coding Strategy of ECOC SVMs Internatonal Journal of Grd and Dstrbuted Computng Vol.7, No. (04), pp.67-76 http://dx.do.org/0.457/jgdc.04.7..07 Performance Analyss and Codng Strategy of ECOC SVMs Zhgang Yan, and Yuanxuan Yang, School

More information

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting Causal, Explanatory Forecastng Assumes cause-and-effect relatonshp between system nputs and ts output Forecastng wth Regresson Analyss Rchard S. Barr Inputs System Cause + Effect Relatonshp The job of

More information

Probabilistic Latent Semantic User Segmentation for Behavioral Targeted Advertising*

Probabilistic Latent Semantic User Segmentation for Behavioral Targeted Advertising* Probablstc Latent Semantc User Segmentaton for Behavoral Targeted Advertsng* Xaohu Wu 1,2, Jun Yan 2, Nng Lu 2, Shucheng Yan 3, Yng Chen 1, Zheng Chen 2 1 Department of Computer Scence Bejng Insttute of

More information

8 Algorithm for Binary Searching in Trees

8 Algorithm for Binary Searching in Trees 8 Algorthm for Bnary Searchng n Trees In ths secton we present our algorthm for bnary searchng n trees. A crucal observaton employed by the algorthm s that ths problem can be effcently solved when the

More information

Searching for Interacting Features for Spam Filtering

Searching for Interacting Features for Spam Filtering Searchng for Interactng Features for Spam Flterng Chuanlang Chen 1, Yun-Chao Gong 2, Rongfang Be 1,, and X. Z. Gao 3 1 Department of Computer Scence, Bejng Normal Unversty, Bejng 100875, Chna 2 Software

More information

An Alternative Way to Measure Private Equity Performance

An Alternative Way to Measure Private Equity Performance An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate

More information

A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION. Michael E. Kuhl Radhamés A. Tolentino-Peña

A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION. Michael E. Kuhl Radhamés A. Tolentino-Peña Proceedngs of the 2008 Wnter Smulaton Conference S. J. Mason, R. R. Hll, L. Mönch, O. Rose, T. Jefferson, J. W. Fowler eds. A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION

More information

Can Auto Liability Insurance Purchases Signal Risk Attitude?

Can Auto Liability Insurance Purchases Signal Risk Attitude? Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159-164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? Chu-Shu L Department of Internatonal Busness, Asa Unversty, Tawan Sheng-Chang

More information

THE APPLICATION OF DATA MINING TECHNIQUES AND MULTIPLE CLASSIFIERS TO MARKETING DECISION

THE APPLICATION OF DATA MINING TECHNIQUES AND MULTIPLE CLASSIFIERS TO MARKETING DECISION Internatonal Journal of Electronc Busness Management, Vol. 3, No. 4, pp. 30-30 (2005) 30 THE APPLICATION OF DATA MINING TECHNIQUES AND MULTIPLE CLASSIFIERS TO MARKETING DECISION Yu-Mn Chang *, Yu-Cheh

More information

Detecting Credit Card Fraud using Periodic Features

Detecting Credit Card Fraud using Periodic Features Detectng Credt Card Fraud usng Perodc Features Alejandro Correa Bahnsen, Djamla Aouada, Aleksandar Stojanovc and Björn Ottersten Interdscplnary Centre for Securty, Relablty and Trust Unversty of Luxembourg,

More information

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College Feature selecton for ntruson detecton Slobodan Petrovć NISlab, Gjøvk Unversty College Contents The feature selecton problem Intruson detecton Traffc features relevant for IDS The CFS measure The mrmr measure

More information

Calculating the high frequency transmission line parameters of power cables

Calculating the high frequency transmission line parameters of power cables < ' Calculatng the hgh frequency transmsson lne parameters of power cables Authors: Dr. John Dcknson, Laboratory Servces Manager, N 0 RW E B Communcatons Mr. Peter J. Ncholson, Project Assgnment Manager,

More information

Gender Classification for Real-Time Audience Analysis System

Gender Classification for Real-Time Audience Analysis System Gender Classfcaton for Real-Tme Audence Analyss System Vladmr Khryashchev, Lev Shmaglt, Andrey Shemyakov, Anton Lebedev Yaroslavl State Unversty Yaroslavl, Russa [email protected], [email protected], [email protected],

More information

Stochastic Protocol Modeling for Anomaly Based Network Intrusion Detection

Stochastic Protocol Modeling for Anomaly Based Network Intrusion Detection Stochastc Protocol Modelng for Anomaly Based Network Intruson Detecton Juan M. Estevez-Tapador, Pedro Garca-Teodoro, and Jesus E. Daz-Verdejo Department of Electroncs and Computer Technology Unversty of

More information

Latent Class Regression. Statistics for Psychosocial Research II: Structural Models December 4 and 6, 2006

Latent Class Regression. Statistics for Psychosocial Research II: Structural Models December 4 and 6, 2006 Latent Class Regresson Statstcs for Psychosocal Research II: Structural Models December 4 and 6, 2006 Latent Class Regresson (LCR) What s t and when do we use t? Recall the standard latent class model

More information

Using Supervised Clustering Technique to Classify Received Messages in 137 Call Center of Tehran City Council

Using Supervised Clustering Technique to Classify Received Messages in 137 Call Center of Tehran City Council Usng Supervsed Clusterng Technque to Classfy Receved Messages n 137 Call Center of Tehran Cty Councl Mahdyeh Haghr 1*, Hamd Hassanpour 2 (1) Informaton Technology engneerng/e-commerce, Shraz Unversty (2)

More information

How To Know The Components Of Mean Squared Error Of Herarchcal Estmator S

How To Know The Components Of Mean Squared Error Of Herarchcal Estmator S S C H E D A E I N F O R M A T I C A E VOLUME 0 0 On Mean Squared Error of Herarchcal Estmator Stans law Brodowsk Faculty of Physcs, Astronomy, and Appled Computer Scence, Jagellonan Unversty, Reymonta

More information

The Use of Analytics for Claim Fraud Detection Roosevelt C. Mosley, Jr., FCAS, MAAA Nick Kucera Pinnacle Actuarial Resources Inc.

The Use of Analytics for Claim Fraud Detection Roosevelt C. Mosley, Jr., FCAS, MAAA Nick Kucera Pinnacle Actuarial Resources Inc. Paper 1837-2014 The Use of Analytcs for Clam Fraud Detecton Roosevelt C. Mosley, Jr., FCAS, MAAA Nck Kucera Pnnacle Actuaral Resources Inc., Bloomngton, IL ABSTRACT As t has been wdely reported n the nsurance

More information

Georey E. Hinton. University oftoronto. Email: [email protected]. Technical Report CRG-TR-96-1. May 21, 1996 (revised Feb 27, 1997) Abstract

Georey E. Hinton. University oftoronto. Email: zoubin@cs.toronto.edu. Technical Report CRG-TR-96-1. May 21, 1996 (revised Feb 27, 1997) Abstract The EM Algorthm for Mxtures of Factor Analyzers Zoubn Ghahraman Georey E. Hnton Department of Computer Scence Unversty oftoronto 6 Kng's College Road Toronto, Canada M5S A4 Emal: [email protected] Techncal

More information

Brigid Mullany, Ph.D University of North Carolina, Charlotte

Brigid Mullany, Ph.D University of North Carolina, Charlotte Evaluaton And Comparson Of The Dfferent Standards Used To Defne The Postonal Accuracy And Repeatablty Of Numercally Controlled Machnng Center Axes Brgd Mullany, Ph.D Unversty of North Carolna, Charlotte

More information

Research on Transformation Engineering BOM into Manufacturing BOM Based on BOP

Research on Transformation Engineering BOM into Manufacturing BOM Based on BOP Appled Mechancs and Materals Vols 10-12 (2008) pp 99-103 Onlne avalable snce 2007/Dec/06 at wwwscentfcnet (2008) Trans Tech Publcatons, Swtzerland do:104028/wwwscentfcnet/amm10-1299 Research on Transformaton

More information

A Simple Approach to Clustering in Excel

A Simple Approach to Clustering in Excel A Smple Approach to Clusterng n Excel Aravnd H Center for Computatonal Engneerng and Networng Amrta Vshwa Vdyapeetham, Combatore, Inda C Rajgopal Center for Computatonal Engneerng and Networng Amrta Vshwa

More information

FREQUENCY OF OCCURRENCE OF CERTAIN CHEMICAL CLASSES OF GSR FROM VARIOUS AMMUNITION TYPES

FREQUENCY OF OCCURRENCE OF CERTAIN CHEMICAL CLASSES OF GSR FROM VARIOUS AMMUNITION TYPES FREQUENCY OF OCCURRENCE OF CERTAIN CHEMICAL CLASSES OF GSR FROM VARIOUS AMMUNITION TYPES Zuzanna BRO EK-MUCHA, Grzegorz ZADORA, 2 Insttute of Forensc Research, Cracow, Poland 2 Faculty of Chemstry, Jagellonan

More information

An interactive system for structure-based ASCII art creation

An interactive system for structure-based ASCII art creation An nteractve system for structure-based ASCII art creaton Katsunor Myake Henry Johan Tomoyuk Nshta The Unversty of Tokyo Nanyang Technologcal Unversty Abstract Non-Photorealstc Renderng (NPR), whose am

More information

Bayesian Network Based Causal Relationship Identification and Funding Success Prediction in P2P Lending

Bayesian Network Based Causal Relationship Identification and Funding Success Prediction in P2P Lending Proceedngs of 2012 4th Internatonal Conference on Machne Learnng and Computng IPCSIT vol. 25 (2012) (2012) IACSIT Press, Sngapore Bayesan Network Based Causal Relatonshp Identfcaton and Fundng Success

More information

How To Analyze News From A News Report

How To Analyze News From A News Report , pp. 385-396 http://dx.do.org/10.14257/jmue.2014.9.11.37 Topc Sentment Analyss n Chnese News Ouyang Chunpng, Zhou Wen +, Yu Yng, Lu Zhmng and Yang Xaohua School of Computer Scence and Technology, Unversty

More information

Active Learning for Interactive Visualization

Active Learning for Interactive Visualization Actve Learnng for Interactve Vsualzaton Tomoharu Iwata Nel Houlsby Zoubn Ghahraman Unversty of Cambrdge Unversty of Cambrdge Unversty of Cambrdge Abstract Many automatc vsualzaton methods have been. However,

More information

Traffic-light a stress test for life insurance provisions

Traffic-light a stress test for life insurance provisions MEMORANDUM Date 006-09-7 Authors Bengt von Bahr, Göran Ronge Traffc-lght a stress test for lfe nsurance provsons Fnansnspetonen P.O. Box 6750 SE-113 85 Stocholm [Sveavägen 167] Tel +46 8 787 80 00 Fax

More information

Calculation of Sampling Weights

Calculation of Sampling Weights Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a two-stage stratfed cluster desgn. 1 The frst stage conssted of a sample

More information

Robust Design of Public Storage Warehouses. Yeming (Yale) Gong EMLYON Business School

Robust Design of Public Storage Warehouses. Yeming (Yale) Gong EMLYON Business School Robust Desgn of Publc Storage Warehouses Yemng (Yale) Gong EMLYON Busness School Rene de Koster Rotterdam school of management, Erasmus Unversty Abstract We apply robust optmzaton and revenue management

More information

Predicting Software Development Project Outcomes *

Predicting Software Development Project Outcomes * Predctng Software Development Project Outcomes * Rosna Weber, Mchael Waller, June Verner, Wllam Evanco College of Informaton Scence & Technology, Drexel Unversty 3141 Chestnut Street Phladelpha, PA 19104

More information

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification Lecture 4: More classfers and classes C4B Machne Learnng Hlary 20 A. Zsserman Logstc regresson Loss functons revsted Adaboost Loss functons revsted Optmzaton Multple class classfcaton Logstc Regresson

More information

Ring structure of splines on triangulations

Ring structure of splines on triangulations www.oeaw.ac.at Rng structure of splnes on trangulatons N. Vllamzar RICAM-Report 2014-48 www.rcam.oeaw.ac.at RING STRUCTURE OF SPLINES ON TRIANGULATIONS NELLY VILLAMIZAR Introducton For a trangulated regon

More information

The Current Employment Statistics (CES) survey,

The Current Employment Statistics (CES) survey, Busness Brths and Deaths Impact of busness brths and deaths n the payroll survey The CES probablty-based sample redesgn accounts for most busness brth employment through the mputaton of busness deaths,

More information

Detecting Algorithmically Generated Malicious Domain Names

Detecting Algorithmically Generated Malicious Domain Names Detectng Algorthmcally Generated Malcous Doman Names Sandeep Yadav, Ashwath K.K. Reddy, and A.L. Narasmha Reddy Department of Electrcal and Computer Engneerng Texas A&M Unversty College Staton, TX 77843,

More information

BERNSTEIN POLYNOMIALS

BERNSTEIN POLYNOMIALS On-Lne Geometrc Modelng Notes BERNSTEIN POLYNOMIALS Kenneth I. Joy Vsualzaton and Graphcs Research Group Department of Computer Scence Unversty of Calforna, Davs Overvew Polynomals are ncredbly useful

More information

320 The Internatonal Arab Journal of Informaton Technology, Vol. 5, No. 3, July 2008 Comparsons Between Data Clusterng Algorthms Osama Abu Abbas Computer Scence Department, Yarmouk Unversty, Jordan Abstract:

More information

A Secure Password-Authenticated Key Agreement Using Smart Cards

A Secure Password-Authenticated Key Agreement Using Smart Cards A Secure Password-Authentcated Key Agreement Usng Smart Cards Ka Chan 1, Wen-Chung Kuo 2 and Jn-Chou Cheng 3 1 Department of Computer and Informaton Scence, R.O.C. Mltary Academy, Kaohsung 83059, Tawan,

More information

Enterprise Master Patient Index

Enterprise Master Patient Index Enterprse Master Patent Index Healthcare data are captured n many dfferent settngs such as hosptals, clncs, labs, and physcan offces. Accordng to a report by the CDC, patents n the Unted States made an

More information

Title Language Model for Information Retrieval

Title Language Model for Information Retrieval Ttle Language Model for Informaton Retreval Rong Jn Language Technologes Insttute School of Computer Scence Carnege Mellon Unversty Alex G. Hauptmann Computer Scence Department School of Computer Scence

More information

The OC Curve of Attribute Acceptance Plans

The OC Curve of Attribute Acceptance Plans The OC Curve of Attrbute Acceptance Plans The Operatng Characterstc (OC) curve descrbes the probablty of acceptng a lot as a functon of the lot s qualty. Fgure 1 shows a typcal OC Curve. 10 8 6 4 1 3 4

More information

A DATA MINING APPLICATION IN A STUDENT DATABASE

A DATA MINING APPLICATION IN A STUDENT DATABASE JOURNAL OF AERONAUTICS AND SPACE TECHNOLOGIES JULY 005 VOLUME NUMBER (53-57) A DATA MINING APPLICATION IN A STUDENT DATABASE Şenol Zafer ERDOĞAN Maltepe Ünversty Faculty of Engneerng Büyükbakkalköy-Istanbul

More information

Transition Matrix Models of Consumer Credit Ratings

Transition Matrix Models of Consumer Credit Ratings Transton Matrx Models of Consumer Credt Ratngs Abstract Although the corporate credt rsk lterature has many studes modellng the change n the credt rsk of corporate bonds over tme, there s far less analyss

More information

1 Example 1: Axis-aligned rectangles

1 Example 1: Axis-aligned rectangles COS 511: Theoretcal Machne Learnng Lecturer: Rob Schapre Lecture # 6 Scrbe: Aaron Schld February 21, 2013 Last class, we dscussed an analogue for Occam s Razor for nfnte hypothess spaces that, n conjuncton

More information

Implementation of Deutsch's Algorithm Using Mathcad

Implementation of Deutsch's Algorithm Using Mathcad Implementaton of Deutsch's Algorthm Usng Mathcad Frank Roux The followng s a Mathcad mplementaton of Davd Deutsch's quantum computer prototype as presented on pages - n "Machnes, Logc and Quantum Physcs"

More information

A study on the ability of Support Vector Regression and Neural Networks to Forecast Basic Time Series Patterns

A study on the ability of Support Vector Regression and Neural Networks to Forecast Basic Time Series Patterns A study on the ablty of Support Vector Regresson and Neural Networks to Forecast Basc Tme Seres Patterns Sven F. Crone, Jose Guajardo 2, and Rchard Weber 2 Lancaster Unversty, Department of Management

More information

Exploiting Recommendation on Social Media Networks

Exploiting Recommendation on Social Media Networks Internatonal Journal of Scence and Research IJSR) ISSN Onln: 2319-7064 Index Coperncus Value 2013): 6.14 Impact Factor 2013): 4.438 Explotng Recommendaton on Socal Meda Networs Swat A. Adhav 1, Sheetal

More information

Document Clustering Analysis Based on Hybrid PSO+K-means Algorithm

Document Clustering Analysis Based on Hybrid PSO+K-means Algorithm Document Clusterng Analyss Based on Hybrd PSO+K-means Algorthm Xaohu Cu, Thomas E. Potok Appled Software Engneerng Research Group, Computatonal Scences and Engneerng Dvson, Oak Rdge Natonal Laboratory,

More information

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek HE DISRIBUION OF LOAN PORFOLIO VALUE * Oldrch Alfons Vascek he amount of captal necessary to support a portfolo of debt securtes depends on the probablty dstrbuton of the portfolo loss. Consder a portfolo

More information

An Approach to Automatically Constructing Domain Ontology 1

An Approach to Automatically Constructing Domain Ontology 1 An Approach to Automatcally Constructng Doman Ontology 1 Tngtng He 1 2 3 Xaopeng Zhang 1 3 Xnghuo Ye 1 3 1 Department of Computer Scence, Huazhong ormal Unversty 430079 Wuhan, Chna 2 Software College of

More information

How To Calculate The Accountng Perod Of Nequalty

How To Calculate The Accountng Perod Of Nequalty Inequalty and The Accountng Perod Quentn Wodon and Shlomo Ytzha World Ban and Hebrew Unversty September Abstract Income nequalty typcally declnes wth the length of tme taen nto account for measurement.

More information

Data Visualization by Pairwise Distortion Minimization

Data Visualization by Pairwise Distortion Minimization Communcatons n Statstcs, Theory and Methods 34 (6), 005 Data Vsualzaton by Parwse Dstorton Mnmzaton By Marc Sobel, and Longn Jan Lateck* Department of Statstcs and Department of Computer and Informaton

More information

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

How To Understand The Results Of The German Meris Cloud And Water Vapour Product Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPP-ATBD-ClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller

More information

PEER REVIEWER RECOMMENDATION IN ONLINE SOCIAL LEARNING CONTEXT: INTEGRATING INFORMATION OF LEARNERS AND SUBMISSIONS

PEER REVIEWER RECOMMENDATION IN ONLINE SOCIAL LEARNING CONTEXT: INTEGRATING INFORMATION OF LEARNERS AND SUBMISSIONS PEER REVIEWER RECOMMENDATION IN ONLINE SOCIAL LEARNING CONTEXT: INTEGRATING INFORMATION OF LEARNERS AND SUBMISSIONS Yunhong Xu, Faculty of Management and Economcs, Kunmng Unversty of Scence and Technology,

More information

SIMPLE LINEAR CORRELATION

SIMPLE LINEAR CORRELATION SIMPLE LINEAR CORRELATION Smple lnear correlaton s a measure of the degree to whch two varables vary together, or a measure of the ntensty of the assocaton between two varables. Correlaton often s abused.

More information

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network *

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network * JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 24, 819-840 (2008) Data Broadcast on a Mult-System Heterogeneous Overlayed Wreless Network * Department of Computer Scence Natonal Chao Tung Unversty Hsnchu,

More information

Fast Fuzzy Clustering of Web Page Collections

Fast Fuzzy Clustering of Web Page Collections Fast Fuzzy Clusterng of Web Page Collectons Chrstan Borgelt and Andreas Nürnberger Dept. of Knowledge Processng and Language Engneerng Otto-von-Guercke-Unversty of Magdeburg Unverstätsplatz, D-396 Magdeburg,

More information

Detecting Global Motion Patterns in Complex Videos

Detecting Global Motion Patterns in Complex Videos Detectng Global Moton Patterns n Complex Vdeos Mn Hu, Saad Al, Mubarak Shah Computer Vson Lab, Unversty of Central Florda {mhu,sal,shah}@eecs.ucf.edu Abstract Learnng domnant moton patterns or actvtes

More information

+ + + - - This circuit than can be reduced to a planar circuit

+ + + - - This circuit than can be reduced to a planar circuit MeshCurrent Method The meshcurrent s analog of the nodeoltage method. We sole for a new set of arables, mesh currents, that automatcally satsfy KCLs. As such, meshcurrent method reduces crcut soluton to

More information

ECE544NA Final Project: Robust Machine Learning Hardware via Classifier Ensemble

ECE544NA Final Project: Robust Machine Learning Hardware via Classifier Ensemble 1 ECE544NA Fnal Project: Robust Machne Learnng Hardware va Classfer Ensemble Sa Zhang, [email protected] Dept. of Electr. & Comput. Eng., Unv. of Illnos at Urbana-Champagn, Urbana, IL, USA Abstract In

More information

Extending Probabilistic Dynamic Epistemic Logic

Extending Probabilistic Dynamic Epistemic Logic Extendng Probablstc Dynamc Epstemc Logc Joshua Sack May 29, 2008 Probablty Space Defnton A probablty space s a tuple (S, A, µ), where 1 S s a set called the sample space. 2 A P(S) s a σ-algebra: a set

More information

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems

Joint Scheduling of Processing and Shuffle Phases in MapReduce Systems Jont Schedulng of Processng and Shuffle Phases n MapReduce Systems Fangfe Chen, Mural Kodalam, T. V. Lakshman Department of Computer Scence and Engneerng, The Penn State Unversty Bell Laboratores, Alcatel-Lucent

More information