Clustering. Cluster Analysis 群聚分析. The K-Means Clustering Method. Clustering 的一般應用. Example. Pattern Recognition 圖樣識別. Cluster 群聚 : 一群 data objects
|
|
- Sharlene McLaughlin
- 5 years ago
- Views:
Transcription
1 The K-Means Clusterng Method Eample Clusterng Assgn each obects to most smlar center reassgn Update the cluster means reassgn K= Arbtrarly choose K obect as ntal cluster center Update the cluster means Cluster Analyss 群聚分析 Clusterng 的一般應用 Cluster 群聚 : 一群 data obects Pattern Recognton 圖樣識別 在同一群內相當相似 在不同群內非常不相似 Cluster analyss 把資料依相似性分群 Clusterng 是 unsupervsed classfcaton: 無預先設好的類別標籤 Spatal Data Analyss 空間資料分析 create thematc maps n GIS by clusterng feature spaces detect spatal clusters and eplan them n spatal data mnng Image Processng 影像處理 Economc Scence (especally market research) Typcal applcatons 作為了解資料分佈的工具 (stand-alone tool) 作為其他方法的 preprocessng step WWW Document classfcaton Cluster Weblog data to dscover groups of smlar access patterns
2 Eamples of Clusterng Applcatons 何謂 Good Clusterng? Marketng: Help marketers dscover dstnct groups n ther customer bases, and then use ths knowledge to develop targeted marketng programs Land use: Identfcaton of areas of smlar land use n an earth observaton database Insurance: Identfyng groups of motor nsurance polcy holders wth a hgh average clam cost Cty-plannng: Identfyng groups of houses accordng to ther house type, value, and geographcal locaton Earth-quake studes: Observed earth quake epcenters should be clustered along contnent faults 好的分群方法產生高品質的 clusters hgh ntra-class smlarty (cluster 內 : 高相似 ) low nter-class smlarty (cluster 間 : 低相似 ) 結果的品質決定於 clusterng 方法的 smlarty measure mplementaton clusterng 方法的品質也可以用 找出 ( 部分或全部 ) 隱藏的 pattern 能力 來度量 Clusterng 需求 擴充性 (scalablty ) 處理各種型態的屬性 (types of attrbutes) 找出任意形狀的 cluster 決定輸入參數時需盡量減少所需的 doman knowledge 處理 nose 及 outler 的能力 對輸入資料的順序要 nsenstve hgh dmensonalty 可以整合 user-specfed constrants Interpretablty usablty Data matr (two modes) n ob. * p var. 資料結構 Dssmlarty matr (one mode) n * n n d(,) d(, ) : d ( n,) f f nf d (,) : d ( n,) : p p np
3 Clusterng Qualty 的度量 Dssmlarty/Smlarty metrc: 以 dstance functon 表示,d(, ) Dstance functons 的定義依照變數型態而不同 nterval-scaled, boolean, categorcal, ordnal and rato varables 依照各個應用與資料的意義訂定變數的 weghts 有時很難定義 smlar enough or good enough 答案很主觀 Clusterng Analyss 的資料型態 Interval-scaled varables weght, heght, lattude, (roughly lnear) Bnary varables symmetrc: gender asymmetrc: fever (Y/N), test (P/N) Nomnal, ordnal, and rato varables map_color, weather; orderng; Ae Bt Varables of med types Interval-valued varables Smlarty and Dssmlarty Between Obects Standardze data 先標準化 ( f 變成 z f ) 算 mean absolute devaton Dstances: 度量兩 data obects 的 smlarty 或 dssmlarty where s = ( f n n m + f f f f nf f m = ( f f f nf 算 standardzed measurement (z-score) m f f z = f s 用 mean absolute devaton 比用 standard devaton 更 robust standard devaton 對差值平方 f m + + ). m ) m f : 平均 propertes d(,) d(,) = d(,) = d(,) d(,) d(,k) + d(k,) Manhattan dstance Eucldean dstance d(, ) = p p d(, ) = ( p p )
4 Smlarty and Dssmlarty (Cont.) Bnary Varables Mnkowsk dstance: q q q d (, ) = q ( ) p p = (,,, p ) and = (,,, p ) 是兩個 p- dmensonal 的 data obects, q 是正整數 A contngency table for bnary data 各種可能 Obect Obect sum a b a+ b c d c+ d sum a+ c b+ d p Manhattan dstance: q = Eucldean dstance: q = Smple matchng coeffcent (nvarant, f the bnary varable s symmetrc): d (, ) = b + c a + b + c + d Jaccard coeffcent (nonnvarant f the bnary varable s asymmetrc): d (, ) = b c a + + b + c Dssmlarty between Bnary Varables Nomnal Varables Eample gender 是 symmetrc attrbute 其他是 asymmetrc bnary attrbute 讓 Y 跟 P 為, N 為 Name Gender Fever Cough Test- Test- Test- Test- Jack M Y N P N N N Mary F Y N P N P N Jm M Y P N N N N + d ( ack, mary ) = = d ( ack, m ) = = d ( m, mary ) = =. + + bnary varable 的 generalzaton : 超過兩種狀態, 如 red, yellow, blue, green Method : Smple matchng m: # of matches, p: total # of varables Method : use a large number of bnary varables creatng a new bnary varable for each of the M nomnal states Show an eample d (, ) = p p m
5 Ordnal Varables Rato-Scaled Varables Can be dscrete or contnuous Order s mportant, e.g., rank Can be treated lke nterval-scaled 將 f 用他的 rank 替代, r f {,, M f } 將各個 ordnal varable 對應到 [, ] 取代 -th obect 的 f-th 變數 r f z = f M f Rato-scaled varable: a postve measurement on a nonlnear scale, appromately at eponental scale, such as Ae Bt or Ae -Bt Methods: treat them lke nterval-scaled varables not a good choce! (why? the scale can be dstorted) apply logarthmc transformaton, (log-log maybe) 用 nterval-scaled varables 的方法計算 dssmlarty y f = log( f ) Eample treat them as contnuous ordnal data as ordnal data, treat ther rank as nterval-scaled Varables of Med Types Maor Clusterng Approaches 資料庫可能同時包含這六種 varables symmetrc bnary, asymmetrc bnary, nomnal, ordnal, nterval and rato 可以用 weghted formula 來結合 Σ d (, ) = ( f ) ( f ) = p ( f ) δ f = 當 f 是 bnary or nomnal: d (f) = f f = f, otherwse d (f) = 當 f 是 nterval-based: d (f) = f - f /ma( f )-mn( f ) 當 f 是 ordnal or rato-scaled 算 ranks r f 把 z f 當 nterval-scaled p f Σ δ d z f = r f M f δ (f) = : () 缺 f 或 f () f = f =, f asymmetrc 其他 : δ (f) = Parttonng algorthms Construct varous parttons evaluatethem by some crteron Herarchyalgorthms Create a herarchcal decomposton of the set of data (or obects) usng some crteron Densty-based: based on connectvty and densty functons Grd-based: based on a multple-level granularty structure Model-based: A model s hypotheszed for each of the clusters and the dea s to fnd the best ft of that model to each other
6 Parttonng Algorthms: Basc Concept The K-Means Clusterng Method Parttonng method: Construct a partton of a database D of n obects nto a set of k clusters Gven k, the k-means algorthm s mplemented n four steps: Gven a k, fnd a partton of k clusters that optmzes the chosen parttonng crteron Global optmal: ehaustvely enumerate all parttons Heurstc methods: k-means and k-medods algorthms Partton obects nto k nonempty subsets Compute seed ponts as the centrods of the clusters of the current partton (the centrod s the center,.e., mean pont, of the cluster) k-means (MacQueen ): Each cluster s represented by the center of the cluster k-medods or PAM (Partton around medods) (Kaufman & Rousseeuw ): Each cluster s represented by one of the obects n the cluster Assgn each obect to the cluster wth the nearest seed pont Go back to Step, stop when no more new assgnment The K-Means Clusterng Method Comments on the K-Means Method Eample Strength: Relatvely effcent: O(tkn), where n s # obects, k s # clusters, and t s # teratons. Normally, k, t << n. K= Arbtrarly choose K obect as ntal cluster center Assgn each obects to most smlar center reassgn Update the cluster means Update the cluster means reassgn Comparng: PAM: O(k(n-k) ), CLARA: O(ks + k(n-k)) Comment: Often termnates at a local optmum. Weakness Applcable only when mean s defned, then what about categorcal data? Need to specfy k, the number of clusters, n advance Unable to handle nosy data and outlers Not sutableto dscover clusters wth non-conve shapes
7 K-Means 的問題 The k-means algorthm s senstve to outlers! Snce an obect wth an etremely large value may substantally dstort the dstrbuton of the data. K-Medods: Instead of takng the mean value of the obect n a cluster as a reference pont, medods can be used, whch s the most centrally located obect n a cluster. Herarchcal Clusterng Use dstance matr as clusterng crtera. Ths method does not requre the number of clusters k as an nput, but needs a termnaton condton Step Step Step Step Step PAM (Parttonng Around Medods, ) Step Step Step Step Step a b c d e a b d e c d e a b c d e agglomeratve (AGNES) dvsve (DIANA) AGNES (Agglomeratve Nestng) Implemented n Splus (e.g) Use Sngle-Lnkage method and the dssmlarty matr Merge nodes that have the least dssmlarty Go on n a non-descendng fashon Eventually all nodes belong to the same cluster Sngle-lnkage: cloest par Complete-lnkage: dstant Dendrogram Shows Herarchcal Clusterng dendrogram: 將 data obects decompose 為數層 nested parttonng (tree of clusters) clusters
8 DIANA (Dvsve Analyss) Implemented n statstcal analyss packages, e.g., Splus Inverse order of AGNES Eventually each node forms a cluster on ts own Densty-Based Clusterng Methods Clusterng based on densty (local cluster crteron), such as densty-connected ponts Maor features: Dscover clusters of arbtrary shape Handle nose One scan Need densty parameters as termnaton condton Several nterestng studes: DBSCAN: Ester, et al. (KDD ) OPTICS: Ankerst, et al (SIGMOD ). DENCLUE: Hnneburg & D. Kem (KDD ) CLIQUE: Agrawal, et al. (SIGMOD ) Model-Based Clusterng Methods Attempt to optmze the ft between the data and some mathematcal model Model-Based Clusterng Methods Statstcal approach Conceptual clusterng COBWEB(Fsher ) AI approach a prototype for each cluster (called eemplar) put new ob. to the most smlar eemplar Neural network approach Self-Organzaton feature Map (SOM) several unts competng for the current obect
9 Self-organzng feature maps (SOMs) What Is Outler Dscovery? Clusterng s also performed by havng several unts competng for the current obect The unt whose weght vector s closest to the current obect wns The wnner and ts neghbors learn by havng ther weghts adusted Useful for vsualzng hgh-dmensonal data n - or -D space Eample Tool 何謂 outlers? Mchael Jordon CEO 薪水 age = 那些跟其他資料相當不相似的資料 (consderably dssmlar!) Problem:fnd top k outlers among n obects Applcatons: Credt card/ Telecom fraud detecton Customer segmentaton Medcal analyss Approaches Statstcal-based Dstance-based Devaton-based Outler Dscovery: Statstcal Approaches Dstance-Based Approach 參數 : p ( 為一分數 ), D Dstance-based outler DB(p, D)-outler: dataset S 中的 obect O, S 中至少 p 的 obect 跟 O 的距離大於 D 沒有夠多的鄰居 Assume a model underlyng dstrbuton that generates data set (e.g. normal dstrbuton) Use dscordancy tests dependng on data dstrbuton dstrbuton parameter (e.g., mean, varance) number of epected outlers Drawbacks most tests are for sngle attrbute In many cases, data dstrbuton may not be known dstance-based outler mnng algorthms Inde-based algorthm Nested-loop algorthm Cell-based algorthm
10 Summary Constrant-Based Clusterng Cluster analyss groups obects based on ther smlarty cluster analyss has wde applcatons Measure of smlarty can be computed for varous types of data Clusterng algorthms can be categorzed nto parttonng methods herarchcal methods densty-based methods grd-based methods model-based methods Outler detecton and analyss useful for fraud detecton, etc. performed by statstcal, dstance-based or devaton-based approaches research ssues: constrant-based clusterng ATM allocaton problem
Clustering. A. Bellaachia Page: 1
Clusterng. Obectves.. Clusterng.... Defntons... General Applcatons.3. What s a good clusterng?. 3.4. Requrements 3 3. Data Structures 4 4. Smlarty Measures. 4 4.. Standardze data.. 5 4.. Bnary varables..
More informationMachine Learning: Algorithms and Applications
14/05/1 Machne Learnng: Algorthms and Applcatons Florano Zn Free Unversty of Bozen-Bolzano Faculty of Computer Scence Academc Year 011-01 Lecture 10: 14 May 01 Unsupervsed Learnng cont Sldes courtesy of
More informationChapter 7. Digital Arithmetic and Arithmetic Circuits. Signed/Unsigned Binary Numbers
Chapter 7 Digital Arithmetic and Arithmetic Circuits Signed/Unsigned Binary Numbers Signed Binary Number: A binary number of fixed length whose sign (+/ ) is represented by one bit (usually MSB) and its
More information描述性資料採礦 Descriptive Data Mining
描述性資料採礦 Descriptive Data Mining 李御璽 (Yue-Shi Lee) 銘傳大學資訊工程學系 leeys@mail.mcu.edu.tw Agenda Cluster Analysis ( 集群分析 ) 找出資料間的內部結構 Association Rules ( 關聯規則 ) 找出那些事件常常一起出現 Sequence Clustering ( 時序群集 ) Clustering
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 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 informationChapter 7. Signed/Unsigned Binary Numbers. Digital Arithmetic and Arithmetic Circuits. Unsigned Binary Arithmetic. Basic Rules (Unsigned)
Chapter 7 Digital rithmetic and rithmetic Circuits igned/unsigned inary Numbers igned inary Number: binary number of fixed length whose sign (+/ ) is represented by one bit (usually M) and its magnitude
More informationSurvey of Cluster Analysis and its Various Aspects
Harmnder Kaur et al, Internatonal Journal of Computer Scence and Moble Computng, Vol.4 Issue.0, October- 05, pg. 353-363 Avalable Onlne at www.csmc.com Internatonal Journal of Computer Scence and Moble
More information全面強化電路設計與模擬驗證. Addi Lin / Graser 2 / Sep / 2016
全面強化電路設計與模擬驗證 Addi Lin / Graser 2 / Sep / 2016 Agenda OrCAD Design Solution OrCAD Capture 功能應用 OrCAD Capture CIS 介紹 OrCAD PSpice 模擬與驗證 OrCAD Design Solution Powerful and Widely Used Design Solution Front-to-Back
More informationUnsupervised Learning and Clustering
Unsupervsed Learnng and Clusterng Why consder unlabeled samples?. Collectng and labelng large set of samples s costly Gettng recorded speech s free, labelng s tme consumng 2. Classfer could be desgned
More informationHierarchical clustering for gene expression data analysis
Herarchcal clusterng for gene expresson data analyss Gorgo Valentn e-mal: valentn@ds.unm.t Clusterng of Mcroarray Data. Clusterng of gene expresson profles (rows) => dscovery of co-regulated and functonally
More informationFigure 1 Microsoft Visio
Pattern-Oriented Software Design (Fall 2013) Homework #1 (Due: 09/25/2013) 1. Introduction Entity relation (ER) diagrams are graphical representations of data models of relation databases. In the Unified
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 information港專單一登入系統 (SSO) 讓本校的同學, 全日制及兼職老師只要一個登入帳戶, 便可同時使用由本校提供的網上系統及服務, 包括 Blackboard 網上學習平台, 港專電郵服務, 圖書館電子資料庫及其他教學行政系統.
港專單一登入系統 (SSO) 讓本校的同學, 全日制及兼職老師只要一個登入帳戶, 便可同時使用由本校提供的網上系統及服務, 包括 Blackboard 網上學習平台, 港專電郵服務, 圖書館電子資料庫及其他教學行政系統. 港專單一登入網站網址 http://portal.hkct.edu.hk (HKCT 之教職員, 學生 ) http://portal.ctihe.edu.hk (CTIHE 之教職員,
More informationJAVA Programming Language Homework V: Overall Review
JAVA Programming Language Homework V: Overall Review ID: Name: 1. Given the following Java code: [5 points] 1. public class SimpleCalc { 2. public int value; 3. public void calculate(){ value = value +
More informationFrame Relay 訊框中繼 FRSW S0/0 S0/1
Frame Relay 訊框中繼 將路由器設定為訊框中繼交換器以進行 frame relay 實驗 : 首先練習設定兩個埠的 frame relay switch FRSW S0/0 S0/1 介面 S0/0 介面 S0/1 102 201 DLI 102 DLI 201 Router(config)# hostname FRSW FRSW(config)# frame-relay switching
More informationPC Link Mode. Terminate PC Link? Esc. [GO]/[Esc] - - [GO]/[Esc] 轉接座未放滿. Make auto accord with socket mounted? [GO]/[Esc] Copy to SSD E0000
Start SU-6808 EMMC Programmer V.0bd7 [ ]Link PC / [ ]Menu [ ] >.Select project.make new project.engineer mode.reset counter 5.Link to PC [ ] PC disconnected PC connected Select project SEM0G9C_A.prj Terminate
More informationMP3 Codec Design 吳炳飛教授. Chaotic Systems & Signal Processing Lab, CSSP Lab. CSSP Lab:
MP3 Codec Design 吳炳飛教授 國立交通大學 電機與控制工程學系 CSSP Lab: http://cssp.cn.nctu.edu.tw Chaotic Systems & Signal Processing Lab, CSSP Lab July 5, 2004 Chapter 1 Introduction to MP3 Chapter 1: Introduction to MP3
More informationRA8835. Dot Matrix LCD Controller Q&A. Preliminary Version 1.2. July 13, RAiO Technology Inc.
RAiO Dot Matrix LCD Controller Q&A Preliminary Version 1.2 July 13, 2009 RAiO Technology Inc. Copyright RAiO Technology Inc. 2009 Update History Version Date Description 1.0 July 13, 2009 Preliminary Version
More informationOracle Database 11g Overview
Oracle Database 11g Overview Charlie 廖志華倍力資訊資深系統顧問 Great Year for Oracle Database Database Market Database for SAP 14.3% 48.6% 9% 3% 17% 4% 15.0% 22.0% 67% Oracle IBM Microsoft Other
More informationSyntest Tool 使用說明. Speaker: Yu-Hsien Cheng Adviser: Kuen-Jong Lee. VLSI/CAD Training Course
Syntest Tool 使用說明 Speaker: Yu-Hsien Cheng Adviser: Kuen-Jong Lee yhc97@beethoven.ee.ncku.edu.tw VLSI/CAD Training Course Foreword Why testing? Class.2 Why Testing? Economics! Reduce test cost (enhance
More information2009 OB Workshop: Structural Equation Modeling. Changya Hu, Ph.D. NCCU 2009/07/ /07/03
Amos Introduction 2009 OB Workshop: Structural Equation Modeling Changya Hu, Ph.D. NCCU 2009/07/02- 2 Contents Amos Basic Functions Observed Variable Path Analysis Confirmatory Factor Analysis Full Model
More informationSoftware Architecture Case Study: Applying Layer in SyncFree
Software Architecture Case Study: Applying Layer in SyncFree Chien-Tsun Chen Department of Computer Science and Information Engineering National Taipei University of Technology, Taipei 06, Taiwan ctchen@ctchen.idv.tw
More informationDavid M. Kroenke and David J. Auer Database Processing Fundamentals, Design, and Implementation
David M. Kroenke and David J. Auer Database Processing Fundamentals, Design, and Implementation Chapter Six: Transforming Data Models into Database Designs 6-1 Chapter Objectives To understand how to transform
More informationWhat is a Better Program?
軟體的特性 What is a Better Program? 軟體之所謂軟 因為沒有 硬性 不可變 不可挑戰的規則 好處 : 彈性很大, 山不轉路轉, 沒有標準答案, 正常運作就好 C++ Object Oriented Programming 壞處 : 很多小問題合在一起不斷放大, 到處藏污納垢, 沒有標準答案, 不知道到底對了沒有 解決方法 Pei-yih Ting Coding styles
More informationCh. 2: Getting Started
Ch. 2: Getting Started 1 About this lecture Study a few simple algorithms for sorting Insertion Sort Selection Sort, Bubble Sort (Exercises) Merge Sort Show why these algorithms are correct Try to analyze
More informationUse of SCTP for Handoff and Path Selection Strategy in Wireless Network
Use of SCTP for Handoff and Path Selection Strategy in Wireless Network Huai-Hsinh Tsai Grad. Inst. of Networking and Communication Eng., Chaoyang University of Technology s9530615@cyut.edu.tw Lin-Huang
More informationCLAD 考前準備 與 LabVIEW 小技巧
CLAD 考前準備 與 LabVIEW 小技巧 NI 技術行銷工程師 柯璟銘 (Jimmy Ko) jimmy.ko@ni.com LabVIEW 認證 Certified LabVIEW Associate Developer (LabVIEW 基礎認證 ) Certified LabVIEW Associate Developer LabVIEW 全球認證 40 題 (37 題單選,3 題複選
More information用於網頁版權保護的資訊隱藏方法. A Steganographic Method for Copyright Protection of Web Pages
用於網頁版權保護的資訊隱藏方法 A Steganographic Method for Copyright Protection of Web Pages Ya-Hui Chang( 張雅惠 ) and Wen-Hsiang Tsai( 蔡文祥 ) Department of Computer & Information Science National Chiao Tung University
More informationData Mining MTAT (4AP = 6EAP)
Clusterng Data Mnng MTAT018 (AP = 6EAP) Clusterng Jaak Vlo 009 Fall Groupng objects by smlarty Take all data and ask what are typcal examples, groups n data Jaak Vlo and other authors UT: Data Mnng 009
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 informationSPI 功能使用方法 Application Note
1 適用產品 :SM59R16A2 / SM59R08A2 2 SPI 使用概述 : SPI 通信使用 4 個引腳, 分別為 SPI_: 當 master 時資料輸出 ; 當 slave 時資料輸入 SPI_: 當 master 時資料輸入 ; 當 slave 時資料輸出 SPI_SCK: SPI 的時脈信號由 master 主控產生 ; 資料 ( 輸出及輸入 ) 和時脈同步 SPI_SS: 此引腳功能唯有當作
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 informationNCCU 資訊碩專班 Advanced Programming Languages
NCCU 資訊碩專班 Advanced Programming Languages 高等程式語言 Instructor: 資科系陳恭副教授 Spring 2006 Lecture 5: Variables, Assignment, Block, Store More Semantic Concepts Variable Model Binding Concept Blocks and Scopes
More informationAllegro SPB V16 Advance
Allegro SPB V16 Advance Allegro SPB 16.2 Advance Import Logic Back Annotate Netlist Compare Advanced Placement Constraint Management Differential Pair Import Logic Other Cadence Import Logic Other 利用 Other
More information游家德 Jade Freeman 群智信息 / 敦群數位資深架構顧問
游家德 Jade Freeman 群智信息 / 敦群數位資深架構顧問 搜尋對企業的需求方案關係 微軟全面性的搜尋方案及應用價值 Enterprise Search 的基本架構 Microsoft Search Solution 物件模型與客製開發 Microsoft Search Solution 應用與案例 Q&A 每人每天會花 10 分鐘在找企業所需文件, 且還可能找不到! 整合的資料大都雜亂無章,
More informationChapter 1 Introduction to Computers, the Internet, and the Web
Chapter 1 Introduction to Computers, the Internet, and the Web Java technologies are classified into three editions: 1. Standard (J2SE technology) 2. Micro (J2ME technology) 3. Enterprise (J2EE technology)
More informationC B A B B C C C C A B B A B C D A D D A A B D C C D D A B D A D C D B D A C A B
高雄市立右昌國中 106 學年度第二學期第二次段考三年級考科答案 國文科 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. C B D C A C B A D B 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. D C B A D C A B D B 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. C B D C B B C
More informationFrom Suffix Trie to Suffix Tree
Outline Exact String Matching Suffix tree an extremely powerful data structure for string algorithms Input: P and S. Output: All occurrences of P in S. Time: O( P + S ) Technique: Z values of PS. Z(i +
More informationData Mining: Concepts and Techniques. Chapter March 8, 2007 Data Mining: Concepts and Techniques 1
Data Mining: Concepts and Techniques Chapter 7.1-4 March 8, 2007 Data Mining: Concepts and Techniques 1 1. What is Cluster Analysis? 2. Types of Data in Cluster Analysis Chapter 7 Cluster Analysis 3. A
More informationOutline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like:
Self-Organzng Maps (SOM) Turgay İBRİKÇİ, PhD. Outlne Introducton Structures of SOM SOM Archtecture Neghborhoods SOM Algorthm Examples Summary 1 2 Unsupervsed Hebban Learnng US Hebban Learnng, Cntd 3 A
More informationChapter 4 (Part IV) The Processor: Datapath and Control (Parallelism and ILP)
Chapter 4 (Part IV) The Processor: Datapath and Control (Parallelism and ILP) 陳瑞奇 (J.C. Chen) 亞洲大學資訊工程學系 Adapted from class notes by Prof. M.J. Irwin, PSU and Prof. D. Patterson, UCB 4.10 Instruction-Level
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 informationA Deflected Grid-based Algorithm for Clustering Analysis
A Deflected Grd-based Algorthm for Clusterng Analyss NANCY P. LIN, CHUNG-I CHANG, HAO-EN CHUEH, HUNG-JEN CHEN, WEI-HUA HAO Department of Computer Scence and Informaton Engneerng Tamkang Unversty 5 Yng-chuan
More informationEZCast Docking Station
EZCast Docking Station Quick Start Guide Rev. 2.00 Introduction Thanks for choosing EZCast! The EZCast Docking Station contains the cutting-edge EZCast technology, and firmware upgrade will be provided
More informationPassword Protection 此篇文章說明如何在程式中加入密碼保護的機制, 當程式開啟, 使用者必須先輸入使用者帳號及密碼, 若是合法使用者才能進入應用程式
Password Protection 此篇文章說明如何在程式中加入密碼保護的機制, 當程式開啟, 使用者必須先輸入使用者帳號及密碼, 若是合法使用者才能進入應用程式 Step 1. 使用 Visual C++ 6.0 產生一個 MFC Application 1) Project name: PasswordProtection 2) Project type: MFC AppWizard(exe)
More informationGPIB 儀器控制之概念及軟硬體介紹 研華股份有限公司 工業自動化事業群
GPIB 儀器控制之概念及軟硬體介紹 研華股份有限公司 工業自動化事業群 Outline 1. Introduction to Adavntech GPIB Card 2. Introduction to IEEE 488.1 3. Introduction to IEEE 488.2 & SCPI GPIB History General Purpose Interface Bus 由 HP 於
More informationUNIX Basics + shell commands. Michael Tsai 2017/03/06
UNIX Basics + shell commands Michael Tsai 2017/03/06 Reading: http://www.faqs.org/docs/artu/ch02s01.html Where UNIX started Ken Thompson & Dennis Ritchie Multics OS project (1960s) @ Bell Labs UNIX on
More informationEZCast Wire User s Manual
EZCast Wire User s Manual Rev. 2.01 Introduction Thanks for choosing EZCast! The EZCast Wire contains the cutting-edge EZCast technology, and firmware upgrade will be provided accordingly in order to compatible
More information多元化資料中心 的保護策略 技術顧問 陳力維
多元化資料中心 的保護策略 技術顧問 陳力維 現代化的資料保護架構 使用者自助服務 任何儲存設備 影響低 多種還原點選擇 (RPO) Application Server 完整全面的雲端整合 Network Disk Target 容易操作與深入各層的報表能力 管理快照與複製能力 Primary Storage 快速 可靠的還原 (RTO) 完整的磁帶 & 複製管理 單一整合的解決方案 企業級的擴充性
More informationIncrease Productivity and Quality by New Layout Flow
Increase Productivity and Quality by New Layout Flow Jonathan / Graser 16 / Oct / 2015 Design Process Introduction CONSTRAINTS PLACEMENT FANOUT BREAKOUT ROUTING DELAY (ATE) NET-GROUP Topology & Delay Physical
More informationBTC, EMPREX Wireless Keybaord +Mouse + USB dongle. 6309URF III Quick Installation Guide
BTC, EMPREX 6309URF III Quick Installation Guide Hardware Installation 1. Plug the dongle receiver connector into your available USB port on PC. 2. Make sure the batteries of the keyboard and mouse are
More informationEZCast Wire. User s Manual. Rev. 2.00
EZCast Wire User s Manual Rev. 2.00 Introduction Thanks for choosing EZCast! The EZCast Wire contains the cutting-edge EZCast technology, and firmware upgrade will be provided accordingly in order to compatible
More informationUbiquitous Computing Using SIP B 朱文藝 B 周俊男 B 王雋伯
Ubiquitous Computing Using SIP B91902039 朱文藝 B91902069 周俊男 B91902090 王雋伯 Outline Ubiquitous Computing Using SIP 1. Introduction 2. Related Work 3. System Architecture 4. Service Example 1. Introduction
More information桌上電腦及筆記本電腦安裝 Acrobat Reader 應用程式
On a desktop or notebook computer Installing Acrobat Reader to read the course materials The Course Guide, study units and other course materials are provided in PDF format, but to read them you need a
More informationAPPLICATION OF IMPROVED K-MEANS ALGORITHM IN THE DELIVERY LOCATION
An Open Access, Onlne Internatonal Journal Avalable at http://www.cbtech.org/pms.htm 2016 Vol. 6 (2) Aprl-June, pp. 11-17/Sh Research Artcle APPLICATION OF IMPROVED K-MEANS ALGORITHM IN THE DELIVERY LOCATION
More informationDigital imaging & free fall of immersed sphere with wall effects
量測原理與機工實驗 ( 下 ) 熱流實驗 ( 一 ) Digital imaging & free fall of immersed sphere with wall effects May 14-18, 2012 Objective: This week s lab work has two parts: (1) how to record digital video and convert it
More informationK-means and Hierarchical Clustering
Note to other teachers and users of these sldes. Andrew would be delghted f you found ths source materal useful n gvng your own lectures. Feel free to use these sldes verbatm, or to modfy them to ft your
More informationInformation is EVERYTHING 微軟企業混和雲解決方案. November 24, Spenser Lin. Cloud Infra Solution Sales, Microsoft Taiwan
Information is EVERYTHING 微軟企業混和雲解決方案 November 24, 2016 Spenser Lin Cloud Infra Solution Sales, Microsoft Taiwan Value to business Applications and services drive future IT business value Efficiency Innovation
More informationUAK1-C01 USB Interface Data Encryption Lock USB 資料加密鎖. Specifications for Approval
Product Definition C-MING Product Semi-finished Product OEM/ODM Product Component USB Interface Data Encryption Lock USB 資料加密鎖 Specifications for Approval Approval Manager Issued By Revision History Revision
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 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 informationGraph-based Clustering
Graphbased Clusterng Transform the data nto a graph representaton ertces are the data ponts to be clustered Edges are eghted based on smlarty beteen data ponts Graph parttonng Þ Each connected component
More informationOxford isolution. 下載及安裝指南 Download and Installation Guide
Oxford isolution 下載及安裝指南 Download and Installation Guide 系統要求 個人電腦 Microsoft Windows 10 (Mobile 除外 ) Microsoft Windows 8 (RT 除外 ) 或 Microsoft Windows 7 (SP1 或更新版本 ) ( 網上下載 : http://eresources.oupchina.com.hk/oxfordisolution/download/index.html)
More information12/2/2009. Announcements. Parametric / Non-parametric. Case-Based Reasoning. Nearest-Neighbor on Images. Nearest-Neighbor Classification
Introducton to Artfcal Intellgence V22.0472-001 Fall 2009 Lecture 24: Nearest-Neghbors & Support Vector Machnes Rob Fergus Dept of Computer Scence, Courant Insttute, NYU Sldes from Danel Yeung, John DeNero
More informationUnsupervised Learning. Andrea G. B. Tettamanzi I3S Laboratory SPARKS Team
Unsupervised Learning Andrea G. B. Tettamanzi I3S Laboratory SPARKS Team Table of Contents 1)Clustering: Introduction and Basic Concepts 2)An Overview of Popular Clustering Methods 3)Other Unsupervised
More informationJava 程式設計基礎班 (7) 莊坤達台大電信所網路資料庫研究室. Java I/O. Class 7 1. Class 7 2
Java 程式設計基礎班 (7) 莊坤達台大電信所網路資料庫研究室 Email: doug@arbor.ee.ntu.edu.tw Class 7 1 回顧 Java I/O Class 7 2 Java Data Structure 動態資料結構 Grow and shrink at execution time Several types Linked lists Stacks Queues Binary
More informationMultimedia Service Support and Session Management 鍾國麟
Multimedia Service Support and Session Management 鍾國麟 2003-9-31 1 1 Agenda Introduction What is Session? Definition Functions Why need Session Management 2G,Internet,3G SIP Basic Operation User Location
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 information企業無線網路安全解決方案. Eric Wu/ 吳章銘大中國區技術總監. China: Taiwan:
企業無線網路安全解決方案 Eric Wu/ 吳章銘大中國區技術總監 China: +886-1391-0819920 Taiwan: +886-933889776 ewu@arubanetworks.com Agenda Aruba Profile WLAN Solution Evolution RF Site Planning Tools (3+1) RF Optimization Dynamic
More informationJava 程式設計基礎班 (7) 劉根豪台大電機所網路資料庫研究室. Java I/O. Class 7 1. Class 7
Java 程式設計基礎班 (7) 劉根豪台大電機所網路資料庫研究室 Email: kenliu@arbor.ee.ntu.edu.tw 1 回顧 Java I/O 2 1 Java Data Structure 動態資料結構 執行的時候可以動態變大或縮小 類型 Linked lists Stacks Queues Binary trees 3 自我參考類別 (self-referential classes)
More information私有雲公有雲的聯合出擊 領先的運算, 儲存與網路虛擬化技術 靈活的計費模式與經濟性 支援廣大的商業應用場景 涵蓋各類型雲服務 類標準的企業資料中心架構 全球規模與快速部署. 聯合設計的解決方案可為客戶提供最佳的 VMware 和 AWS
私有雲公有雲的聯合出擊 領先的運算, 儲存與網路虛擬化技術 支援廣大的商業應用場景 類標準的企業資料中心架構 靈活的計費模式與經濟性 涵蓋各類型雲服務 全球規模與快速部署 聯合設計的解決方案可為客戶提供最佳的 VMware 和 AWS VMware Cloud on AWS 使用場景 A B C D 雲端遷移資料中心延伸災難備援次世代應用程式 Consolidate Migrate Maintain
More informationInTANK ir2771-s3 ir2772-s3. User Manual
InTANK ir2771-s3 ir2772-s3 User Manual » InTANK...1» InTANK ir2771-s3 & ir2772-s3 產品使用說明... 10 V1.1 Introduction Thank you for purchasing RAIDON products. This manual will introduce the InTANK ir2771-s3
More informationDr. Whai-En Chen. Institute of Computer Science and Information Engineering National Ilan University TEL: #340
IPv6 Transition ( 轉移機制 ) Dr. Whai-En Chen Assistant Professor Institute of Computer Science and Information Engineering National Ilan University wechen@niu.edu.twedu TEL: +886-3-9357400#340 http://alan.ipv6.club.tw
More informationMSI MS-3871 Wireless 11b/g/n + Bluetooth BT2.1 EDR Combo Slim Module. User sguide
MSI MS-3871 Wireless 11b/g/n + Bluetooth BT2.1 EDR Combo Slim Module User sguide i FCC Caution 1. The device complies with Part 15 of the FCC rules. Operation is subject to the following two conditions:
More informationTriangle - Delaunay Triangulator
Triangle - Delaunay Triangulator eryar@163.com Abstract. Triangle is a 2D quality mesh generator and Delaunay triangulator. Triangle was created as part of the Quake project in the school of Computer Science
More informationAnalyzing Popular Clustering Algorithms from Different Viewpoints
1000-9825/2002/13(08)1382-13 2002 Journal of Software Vol.13, No.8 Analyzng Popular Clusterng Algorthms from Dfferent Vewponts QIAN We-nng, ZHOU Ao-yng (Department of Computer Scence, Fudan Unversty, Shangha
More informationClustering algorithms and validity measures
Clusterng algorthms and valdty measures M. Hald, Y. Batstas, M. Vazrganns Department of Informatcs Athens Unversty of Economcs & Busness Emal: {mhal, yanns, mvazrg}@aueb.gr Abstract Clusterng ams at dscoverng
More informationChapter 7 Pointers ( 指標 )
Chapter Pointers ( 指標 ) Outline.1 Introduction.2 Pointer Variable Definitions and Initialization.3 Pointer Operators.4 Calling Functions by Reference.5 Using the const Qualifier with Pointers.6 Bubble
More information臺北巿立大學 104 學年度研究所碩士班入學考試試題
臺北巿立大學 104 學年度研究所碩士班入學考試試題 班別 : 資訊科學系碩士班 ( 資訊科學組 ) 科目 : 計算機概論 ( 含程式設計 ) 考試時間 :90 分鐘 08:30-10:00 總分 :100 分 注意 : 不必抄題, 作答時請將試題題號及答案依照順序寫在答卷上 ; 限用藍色或黑色筆作答, 使用其他顏色或鉛筆作答者, 所考科目以零分計算 ( 於本試題紙上作答者, 不予計分 ) 一 單選題
More informationPreamble Ethernet packet Data FCS
Preamble Ethernet. packet Data FCS Destination Address Source Address EtherType Data ::: Preamble. bytes. Destination Address. bytes. The address(es) are specified for a unicast, multicast (subgroup),
More information一般來說, 安裝 Ubuntu 到 USB 上, 不外乎兩種方式 : 1) 將電腦上的硬碟排線先予以排除, 將 USB 隨身碟插入主機, 以一般光碟安裝方式, 將 Ubuntu 安裝到 USB
Ubuntu 是新一代的 Linux 作業系統, 最重要的是, 它完全免費, 不光是作業系統, 連用軟體都不必錢 為什麼要裝在 USB 隨身碟上? 因為, 你可以把所有的軟體帶著走, 不必在每一台電腦上重新來一次, 不必每一套軟體裝在每一台電腦上都要再一次合法授權 以下安裝方式寫的是安裝完整的 Ubuntu- 企業雲端版本 V. 11.10 的安裝過程, 若是要安裝 Desktop 版本, 由於牽涉到
More informationSSL VPN User Manual (SSL VPN 連線使用手冊 )
SSL VPN User Manual (SSL VPN 連線使用手冊 ) 目錄 前言 (Preface) 1. ACMICPC 2018 VPN 連線說明 -- Pulse Secure for Windows ( 中文版 ):... 2 2. ACMICPC 2018 VPN 連線說明 -- Pulse Secure for Linux ( 中文版 )... 7 3. ACMICPC 2018
More informationCluster Analysis. CSE634 Data Mining
Cluster Analysis CSE634 Data Mining Agenda Introduction Clustering Requirements Data Representation Partitioning Methods K-Means Clustering K-Medoids Clustering Constrained K-Means clustering Introduction
More information國立交通大學 資訊科學與工程研究所 碩士論文 適用於非對稱網路連線之動態用戶的 彈性應用層多點傳播 研究生 : 郭宇軒 指導教授 : 邵家健副教授. Resilient Application Layer Multicast Tailored for
國立交通大學 資訊科學與工程研究所 碩士論文 適用於非對稱網路連線之動態用戶的 彈性應用層多點傳播 Resilient Application Layer Multicast Tailored for Dynamic Peers with Asymmetric Connectivity 研究生 : 郭宇軒 指導教授 : 邵家健副教授 中華民國九十五年七月 適用於非對稱網路連線之動態用戶的彈性應用層多點傳播
More informationWIN Semiconductors. Wireless Information Networking 穩懋半導體 2014 年第四季法人說明會. p.0
WIN Semiconductors Wireless Information Networking 穩懋半導體 2014 年第四季法人說明會 2015 年 3 月 p.0 免責聲明 本資料可能包含對於未來展望的表述 該類表述是基於對現況的 預期, 但同時受限於已知或未知風險或不確定性的影響 因此實 際結果將可能明顯不同於表述內容 除法令要求外, 公司並無義務因應新資訊的產生或未來事件的發生主動更新對未來展望的表述
More informationLotusphere Comes to You 輕鬆打造 Web 2.0 入口網站 IBM Corporation
輕鬆打造 Web 2.0 入口網站 2007 IBM Corporation 議程 Web 2.0 新特性一覽 Web 2.0 入口網站主題開發 用戶端聚合技術 PortalWeb2 主題 開發 AJAX portlets 程式 總結 JSR 286 及 WSRP 2.0 對 AJAX 的支援 AJAX 代理 用戶端 portlet 編程模型 Web 2.0 特性一覽 WP 6.1 提供的 Web
More informationEdConnect and EdDATA
www.hkedcity.net Tryout Programme of Standardised Data Format for e-textbook and e-learning Platform EdConnect and EdDATA 5 December 2018 Agenda Introduction and background Try-out Programme Q&A 電子課本統一數據格式
More informationSystem Programming. System Software: An Introduction to Systems Programming. Leland L. Beck 3rd Edition Addison-Wesley, 1997
System Programming System Software: An Introduction to Systems Programming Leland L. Beck 3rd Edition Addison-Wesley, 1997 1 http://web.thu.edu.tw/ctyang/ 2 http://hpc.csie.thu.edu.tw/ 3 Score List Participation:
More informationHierarchical agglomerative. Cluster Analysis. Christine Siedle Clustering 1
Herarchcal agglomeratve Cluster Analyss Chrstne Sedle 19-3-2004 Clusterng 1 Classfcaton Basc (unconscous & conscous) human strategy to reduce complexty Always based Cluster analyss to fnd or confrm types
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 informationVZ-P18 和 VZ-P38 專業教材提示機 完美的展示效果
VZ-P18 和 VZ-P38 專業教材提示機 完美的展示效果 無與倫比的性能 VZ-P18 和 VZ-P38 專業教材提示機 WolfVision 是一家在全球獲得成功的家族企業, 總部位於歐洲奧地利 WolfVision 在實物提示機和展示解決方案上具有 技術領導者 " 的地位, 在產品品質 技術創新可靠性和易用性方面在全球中樹立了標竿 WolfVision 的 P 系列提示機被認為是市場上最高階的設備
More information计算机组成原理第二讲 第二章 : 运算方法和运算器 数据与文字的表示方法 (1) 整数的表示方法. 授课老师 : 王浩宇
计算机组成原理第二讲 第二章 : 运算方法和运算器 数据与文字的表示方法 (1) 整数的表示方法 授课老师 : 王浩宇 haoyuwang@bupt.edu.cn 1 Today: Bits, Bytes, and Integers Representing information as bits Bit-level manipulations Integers Representation: unsigned
More informationMicrosoft SQL Server 2016 R Services
Microsoft SQL Server 2016 R Services SQL Server 2016: Everything built-in built-in built-in built-in built-in built-in $2,230 80 70 60 50 43 69 49 SQL Server SQL Server SQL Server 40 30 20 10 0 34 29 22
More information報告人 / 主持人 : 林寶樹 Colleges of Computer Science & ECE National Chiao Tung University
行動寬頻尖端技術跨校教學聯盟 - 行動寬頻網路與應用 MiIoT ( Mobile intelligent Internet of Things) 報告人 / 主持人 : 林寶樹 Colleges of Computer Science & ECE National Chiao Tung University Aug 14, 2015 課程簡介 課程綱要 實作平台評估 2 背景說明 目前雲端與行動寬頻緊密結合,
More informationVB 拼圖應用 圖形式按鈕屬性 資科系 林偉川
VB 拼圖應用 資科系 林偉川 圖形式按鈕屬性 Style 屬性 0 ( 標準外觀 ),1( 圖片外觀 ) Picture 屬性 圖形檔案 (VB6) image 屬性 圖形檔案 (VB.NET) Left=Top=0 Width=2052,Height=2052 共有九張圖 1.jpg 9.jpg Form1 執行時視窗為最大化 Windowstate 設為 2 2 1 執行結果 3 path$
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 information第九章結構化查詢語言 SQL - 資料定義語言 (DDL) 資料庫系統設計理論李紹綸著
第九章結構化查詢語言 SQL - 資料定義語言 (DDL) 資料庫系統設計理論李紹綸著 SQL 的資料定義語言 本章內容 建立資料表 修改資料表 刪除資料表 FOREIGN KEY 外鍵條件約束與資料表關聯性 2 資料定義語言可分為下列三種 : SQL 的資料定義語言 CREATE TABLE 指令 : 用來建立一個基底關聯表, 和設定關聯表相關的完整性限制 CREATE VIEW 指令 : 用來建立一個視界,
More informationRelational Database SNOOPYFAMILY 利用 SQL 做查詢 : Select NAME From SNOOPYFAMILY Where SEX = Male ; 結果 : Domains. Primary Key. Cardinality.
資料庫系統實驗室 指導教授 : 張玉盈 1 Relational Database Primary Key SNOOPYFAMILY Male Female Domains ID NAME SEX 利用 SQL 做查詢 : 1 SNOOPY Male 2 CHARLIE BROWN Male 3 SALLY BROWN Female 4 LUCY VAN PELT Female 5 LINUS VAN
More informationLinked Lists. Prof. Michael Tsai 2017/3/14
Linked Lists Prof. Michael Tsai 2017/3/14 What s wrong with Arrays? Inserting a new element 1 3 New 4 42 25 5Empty Deleting an existing element 1 3 42 25 5 Time complexity= O(??) 2 Complexity for the array
More information