Скачать презентацию Data Mining Algorithms for Large-Scale Distributed Systems Presenter Скачать презентацию Data Mining Algorithms for Large-Scale Distributed Systems Presenter

cc23bc1f6236fe9a4d53f65d95e291bc.ppt

  • Количество слайдов: 30

Data Mining Algorithms for Large-Scale Distributed Systems Presenter: Ran Wolff Joint work with Assaf Data Mining Algorithms for Large-Scale Distributed Systems Presenter: Ran Wolff Joint work with Assaf Schuster 2003

What is Data Mining? The automatic analysis of large database The discovery of previously What is Data Mining? The automatic analysis of large database The discovery of previously unknown patterns The generation of a model of the data

Main Data Mining Problems Association rules n Description Classification n Fraud, Churn Clustering n Main Data Mining Problems Association rules n Description Classification n Fraud, Churn Clustering n Analysis He who does this and that will usually do some other thing too These attributes indicate a good behavior - those indicate bad behavior. There are three types of entities

Examples – Classification Customers purchase artifacts in a store Each transaction is described in Examples – Classification Customers purchase artifacts in a store Each transaction is described in terms of a vector of features The owner of the store tries to predict which transactions are fraudulent n n Example: young men who buy small electronics during rash-hours Solution: do not respect checks

Examples – Associations Amazon tracks user queries n Suggests to each user additional books Examples – Associations Amazon tracks user queries n Suggests to each user additional books he would usually be interested in Supermarket finds out “people who buy diapers also buy beer” n Place diapers and beer at opposite sides of the supermarket

Examples – Clustering Resource location n Find the best location for k distribution centers Examples – Clustering Resource location n Find the best location for k distribution centers Feature selection n n Find 1000 concepts which summarize a whole dictionary Extract the meaning out of a document by replacing each work with the appropriate concept w Car for auto, etc.

Why Mine Data of LSD Systems? Data mining is good It is otherwise difficult Why Mine Data of LSD Systems? Data mining is good It is otherwise difficult to monitor an LSD system: lots of data, spread across the system, impossible to collect Many interesting phenomena are inherently distributed (e. g. , DDo. S), it is not enough to just monitor a few nodes

An Example Peers in the Kazza network reveal to the system which files they An Example Peers in the Kazza network reveal to the system which files they have on their disks in exchange to access to the files of their peers The result is a 2 M peers database of people recreational preferences Mining it, you could discover that Matrix fans are also keen of Radio-Head songs n n Promote RH performances in Matrix-Reloaded Ask RH to write the music for Matrix-IV

What is so special about this problem? Huge systems – Huge amounts of data What is so special about this problem? Huge systems – Huge amounts of data Dynamic setting n n System – join / depart Data – constant update Ad-hoc solution Fast convergence

Our Work We developed an association rule mining algorithm that works well in LSD Our Work We developed an association rule mining algorithm that works well in LSD Systems n n n Local and therefore scalable Asynchronous and therefore fast Dynamic and therefore robust Accurate – not approximated Anytime – you get early results fast

In a Teaspoon A distributed data mining algorithm can be described as a series In a Teaspoon A distributed data mining algorithm can be described as a series of distributed decisions Those decisions are reduced to a majority vote We developed a majority voting protocol which has all those good qualities The outcome is an LSD association rule mining (still to come: classification)

Problem Definition – Association Rule Mining (ARM) Problem Definition – Association Rule Mining (ARM)

Solution to Traditional ARM Solution to Traditional ARM

Large-Scale Distributed ARM Large-Scale Distributed ARM

Solution of LSD-ARM No termination Anytime solution Recall Precision Solution of LSD-ARM No termination Anytime solution Recall Precision

Majority Vote in LSD Systems Unknown number of nodes vote 0 or 1 n Majority Vote in LSD Systems Unknown number of nodes vote 0 or 1 n n n Nodes may dynamically change their vote Edges are dynamically added / removed An infra-structure w detects failure w ensures message integrity w maintains a communication forest Each node should decide if the global majority is of 0 or 1

Majority Vote in LSD Systems – cont. Because of the dynamic settings, the algorithm Majority Vote in LSD Systems – cont. Because of the dynamic settings, the algorithm never terminates Instead we measure the percent of correct outputs In static periods that percent ought to converge to 100% In stationary periods we will show it converges to a different percentage n Assume the overall percentage of ones remains the same, but they are constantly switched

LSD-Majority Algorithm Nodes communicates by exchanging messages <s, c> Node u maintains: n n LSD-Majority Algorithm Nodes communicates by exchanging messages Node u maintains: n n n su – its vote, cu – one (for now) – the last it had sent to v – the last it had received from v

LSD-Majority – cont. Node u calculates: n n Captures the current knowledge of u LSD-Majority – cont. Node u calculates: n n Captures the current knowledge of u Captures the current agreement between u and v

LSD-Majority – Rational It is OK if the current knowledge of u is more LSD-Majority – Rational It is OK if the current knowledge of u is more extreme than what it had agreed with v The opposite is not OK n v might assume u supports its decision more strongly than u actually does Tie breaking prefers a negative decision

LSD-Majority – The Protocol LSD-Majority – The Protocol

LSD-Majority – The Protocol The same decision is applied whenever n n n a LSD-Majority – The Protocol The same decision is applied whenever n n n a message is received su changes an edge fails or recovers

LSD-Majority – Example LSD-Majority – Example

LSD-Majority Results LSD-Majority Results

Proof of Correctness Will be given in class Proof of Correctness Will be given in class

Back from Majority to ARM To decide whether an itemset is frequent or not Back from Majority to ARM To decide whether an itemset is frequent or not

Back from Majority to ARM To decide whether a rule is confident or not Back from Majority to ARM To decide whether a rule is confident or not

Additionally Create candidates based on the ad-hoc solution Create rules on-the-fly rather than upon Additionally Create candidates based on the ad-hoc solution Create rules on-the-fly rather than upon termination Our algorithm outputs the correct rules without specifying their global frequency and confidence

Eventual Results By the time the database is scanned once, in parallel, the average Eventual Results By the time the database is scanned once, in parallel, the average node has discovered 95% of the rules, and has less than 10% false rules.