CHAPTER 6 IMPLEMENTATION OF CONVENTIONAL AND INTELLIGENT CLASSIFIER FOR FLAME MONITORING
|
|
- Gwenda Dawson
- 8 years ago
- Views:
Transcription
1 135 CHAPTER 6 IMPLEMENTATION OF CONVENTIONAL AND INTELLIGENT CLASSIFIER FOR FLAME MONITORING 6.1 PROPOSED SETUP FOR FLAME MONITORING IN BOILERS The existing flame monitoring system includes the flame images that are acquired using infrared camera and displayed in the CRT monitor at control room for indicating the presence or absence of flame in the combustion chamber so as to avoid over loading of the furnace which may cause explosion of the boiler. The proposed system has an additional device for transferring the video from the CRT monitor to the computer with various image processing algorithms as shown in Figure 6.1 so as to monitor the combustion conditions. The video of the flame image is acquired using the infrared camera with cooling arrangement. The video is converted into frames using video splitter and these frames are further analyzed. The features are extracted from the flame images and these features are used for flue gases and combustion quality estimation as proposed by Gilabert et al (2005). The intelligent and conventional classifiers are trained with the feature vectors generated from 3 classes of images. The target values for the various intelligent and conventional classifiers are the temperature of
2 136 each group of images, the measured CO values, CO 2, NO x and excess O 2 from the flue gas emissions. The flue gases are measured using gas analyzer. In the test phase, the outputs of proposed algorithms are compared with the measured values of flue gas to decide if any adjustment in the air/fuel ratio is required for the burner. system are The major steps involved in the proposed flame monitoring 1. Infrared camera is placed inside a cooling jacket with servo motor mechanism 2. CCTV set up is placed in the control room 3. TV tuner is installed for transferring the flame video from the CRT monitor on to the Laptop 4. Image processing packages are loaded in the laptop connected to the TV tuner 5. The video file which is split up into frames for further analysis 6. Image processing algorithms for analyzing the constituents of the flame images. 7. Intelligent control strategy to monitor and control the combustion quality 8. The validation of the developed algorithms.
3 137 Figure 6.1 Schematic diagram of the proposed flame monitoring system Data Collection from the Existing Set up The flame images are obtained from the control room of the thermal power plant boiler. Table 6.1 shows the values of the flue gas emissions, flame temperature and combustion quality pertaining to the three classes of flame images as measured from the gas analyzer. Table 6.2 shows 10 samples for each class from the 102 flame images gathered. Of these 51 images are used for training and another 51 images used for testing. The class 1, class 2 and class 3 are referred to as the complete combustion category, partial combustion category and incomplete combustion category respectively. Class 1 (flame1 to flame 18), class 2 (flame 19 to flame 38) and class 3 (flame 39 to flame 51) are of importance for control Engineers to take necessary action. The cropping of each image is chosen to 30 x 30 pixel size, instead any other size. The SOx, CO, NO x, CO 2, % excess O 2, combustion quality and air/fuel ratio measured from the flue gas at the same instant is also recorded from the existing set up.
4 Table 6.1 Measurement data for flue gas emissions corresponding to each class of image Class Combustion conditions Combustion Quality (%) NO x mg/nm 3 CO ppm CO 2 Nm 3 /hr SOx Emissions mg/nm 3 Air/fuel ratio (No units) 1 Complete Ratio is 4:1 890 t/hr-air 2 Partial Combustion 3 Incomplete Combustion 182 t/hr- lignite Ratio is 3:1 600 t/hr air 213 t/hr- lignite Ratio is 2:1 230 t/hr- lignite 400 t/hr - air Flame Temperature in (degree Celsius) Temperature of superheated steam in (degree Celsius) Table 6.2 Sample Flame images Combustion category Class 1 Complete combustion Recorded images Class 2 Partial combustion Class 3 Incomplete combustion 138
5 SOFTWARE USED FOR IMPLEMENTATION OF FLAME MONITORING IN BOILERS The various software packages like MATLAB, Image J and WEKA are used for simulation purpose so as to implement different types of conventional and intelligent schemes Introduction to MATLAB The name MATLAB stands for Matrix Laboratory. The MATLAB is a high performance language for technical computing which integrates computation, visualization, and programming in an easy to use environment where problems and solutions are expressed in familiar mathematical notation. The procedural steps involved in using MATLAB for implementing the flame monitoring system as shown in Figure 6.2. The scientific and engineering graphics application development, including graphical user interface building, the MATLAB is an interactive system whose basic data element is an array that does not require dimensioning. Also this allows solving many technical computing problems, especially those with matrix and vector formulations and in a fraction it would take to write a program in a scalar non interactive language such as C or Formula Translation (FORTRAN). In industry, the MATLAB is the tool of choice for high productivity research, development and analysis. The MATLAB features a family of addon application specific solutions called toolboxes. Most users prefer MATLAB toolbox as its allows learning and applying specialized technology. The toolboxes are comprehensive collections of MATLAB functions (M-files) which extend the MATLAB environment to solve particular classes of problems. The areas in which toolboxes are available include signal processing, image processing, control systems, neural networks, fuzzy logic, wavelets, graphical simulation and many others.
6 140 Figure 6.2 Procedural block diagram for implementation with MATLAB Introduction to WEKA The WEKA workbench contains a collection of visualization tools and algorithms for data analysis and predictive modelling as discussed by Zdravko Markov et al (2005), together with graphical user interfaces for easy access to this functionality. The original non Java version of Weka was a front-end tool for modelling algorithms implemented in other programming languages, plus data preprocessing utilities in C, and a makefile based system for running machine learning experiments. The main user interface in WEKA is the Explorer, but essentially the same functionality can be accessed through the component based Knowledge Flow interface and from the command line. There is also the Experimenter, which allows the systematic comparison of the predictive performance of Weka's machine learning algorithms on a collection of datasets.
7 Overview of WEKA The overview of the WEKA contains the set of tools and the number of modes of operation, each of which includes the following steps and is used for analysis by Pablo Chacon et al (2002). Comprehensive set of tools Preprocessing and data analysis Learning algorithms (for classification, clustering, etc.) Evaluation metrics Three modes of operation Graphical User Interface (GUI) command-line Java Application Programming Interface (API) Modules of WEKA There are four main modulus of WEKA. They are listed as follows Knowledge Flow Explorer Cross Validation Visualization The WEKA reads file in Attribute Relation File Format (ARFF) and its supported attributes are numeric, nominal, string and data types Knowledge Flow Environment (KFL) The KFL is a user friendly tool, developed at University of Waikato in New Zealand. A collection of state-of-art machine learning algorithms and data preprocessing tools are listed as follows;
8 142 Classification, Regression, and clustering Multiple evaluation schemes Feature Selection means choice of right features and data key to successful learning Experimentation Visualization It provides implementation for Regression Classification Clustering Association rules Feature selection 6.3 METHODOLOGY FOR FLAME MONITORING USING IMAGE PROCESSING IN BOILERS The methodology for flame monitoring system involves the following stages in implementation. The first step is the preprocessing which includes denoising, square image extraction and conversion of colour images to gray scale images. The second step includes extraction of the features. These features extracted are used to estimate the classifier performance during training process. The extracted features which yield optimal performance for the classifier are termed as the selected features for testing process. During testing process the selected features are extracted and given as inputs to the classifier for classification based on combustion quality and flue gas emissions of the flame images. The general block diagram for the methodology is shown in Figure 6.3.
9 143 Figure 6.3 Block Diagram for flame monitoring using image processing 6.4 COMBUSTION MONITORING IN POWER STATION BOILERS USING FLD AND RBF NETWORK This section of the research work includes a combination of FLD analysis and a RBF network as proposed by Meng Joo Er et al (2002) for monitoring the combustion conditions for a coal fired boiler to adjust the air/fuel ratio. The overall block diagram for the implementation of the Radial Basis Function network along with Fisher s Linear Discriminant Analysis to identify the flue gas emissions and combustion quality is shown in Figure 6.4. Read image Extract features Apply FLD Train / Test RBF Adjustment of Air/ fuel ratio NO Compare with the threshold YES Maintain existing air/fuel Measure various flue gas and temperature Figure 6.4 Block diagram for flame monitoring using FLD and RBF
10 144 Also, three classes of images corresponding to different burning conditions of the flames have been extracted from continuous video processing. In this, the corresponding temperatures, the carbon monoxide (CO) emissions and those of other flue gases have been obtained through measurement Feature Extraction The feature extraction is most important stage as it serves as the foundation for the classifier for identifying the combustion quality. The flowchart for extracting the features from the captured flame images are given in Figure 6.5 and Figure 6.6. The theory and expressions regarding the FLD algorithm has been discussed in the section The two dimensional flame images obtained from the flame video are preprocessed and the features such as average intensity, area, brightness and orientation, etc., are extracted using standard algorithms. These features are used for faster learning of the various conventional and intelligent classifiers. Figure 6.5 Block diagram to calculate the discriminant vectors Figure 6.6 Flow chart for average intensity, area of high temperature flame and rate area of the high temperature flame
11 Classification using FLD and RBF Network Further, the training and testing of Fisher s Linear Discriminant and Radial Basis Function network (FLDRBF), with the data collected have been carried out and the performance of the algorithms is presented. The classification performance of RBF and FLDRBF is compared with the conventional classification techniques like FLD and EDC. The architecture for RBF is shown in Figure 6.7. Features from the flame images as inputs to RBF Flue gas emissions and combustion quality Input layer Hidden layer Output layer Figure 6.7 Architecture of Radial Basis Function Network 6.5 COMBUSTION QUALITY AND FLUE GAS MONITORING USING PARALLEL ARCHITECTURE OF INTELLIGENT CLASSIFIERS The intelligent classifiers discussed so far are single massive structures which are computationally complex. These single network structures contain a complicated architecture with many number of nodes in the hidden layer or more than one hidden layer itself. Training such networks is tedious as their convergence is not reached at a faster rate. Hence a novel technique based on the combination of Fisher s Linear Discriminant (FLD) analysis with Radial Basis Function Network (RBF) and Back Propagation Algorithm (BPA) for monitoring the combustion conditions of a coal fired
12 146 boiler from the furnace flame so as to adjust the air/fuel ratio is discussed in this section similar to the implementation of hybrid neural networks for slag monitoring in boilers by Tan1et al (2006). The procedure for feature extraction from the flame images is already discussed in section and the same set of features are used for classification purpose. This method includes feature extraction and classification. The training and testing is done using a Parallel architecture of Radial Basis Function network and Back Propagation Algorithm (PRBFBPA) with the previously collected data. The performance of the algorithm is also presented. The images are preprocessed and features are extracted. Training of RBF and BPA was done with 51 images taken from class 1; class 2 and class 3 images and finally the outputs from these networks are combined and given as the input to another RBF so as to obtain the final output. Testing and validation results indicate that PRBFBPA gives maximum classification performance when compared to FLD, RBF and various other combinations of parallel architectures of the neural networks. Classification performance can be improved by further preprocessing of the acquired images. By continuously monitoring the flame images, combustion quality is inferred (complete/partial/incomplete combustion). From the combustion quality the air/fuel ratio can be automatically varied. The following steps are involved in flame video analysis as shown in Figure 6.8. Step 1: Step 2: Step 3: Step 4: Step 5: Step 6: The video file is further processed by splitting into frames using any video splitter The flame images are preprocessed for noise removal The features are extracted Features are reduced Classification using AI techniques Validation of the algorithms with performance measures
13 147 The outputs of BPA and RBF are combined and given as inputs to a second RBF for final classification of the flame images. The block diagram for PRBFBPA is shown in Figure 6.9. Similar structures are used to identify the various flue gas emissions like CO, CO 2, SO x and NO x along with air/fuel ratio and flame temperature. The features obtained from flame images are given as the inputs to the BPA and RBF initially. Now the outputs from BPA and RBF is given to the second RBF network for final classification. The various combination of parallel architectures that are attempted in this work are listed in the Table 6.3(a). The PRBFBPA architecture was found to give good classification among the various combinations of ANN. Now with PRBFBPA the feature combinations are varied to check for the optimal performance. The various combinations of features (three features to BPA with four features to RBF and four features to BPA and three features to RBF) are listed out in the Table 6.3(b). Read image Extract features Apply FLD Train / Test RBF BPA RBF PRBFB Adjustment of Air/ fuel ratio NO Check if the actual output is within tolerance YES Maintain existing air/fuel ratio Measure various flue gas and temperature Figure 6.8 Block diagram for parallel architecture of BPA and RBF
14 148 Figure 6.9 Block diagram for Parallel Radial Basis Function network and Back propagation algorithm Table 6.3(a) Various schemes for the parallel architecture S.No Various combinations for Type of the network parallel architectures Network 1 Network 2 Network 3 1. PRBFBPA BPA RBF RBF 2. Multiple RBF RBF RBF RBF 3. Multiple BPA BPA BPA BPA Table 6.3(b) Various feature combinations for the PRBFBPA Combination S.No Combination 1 Combination 2 Combination 3 Combination 4 Combination 5 Combination 6 Combination 7 Type of the Network Centroid X Centroid Y Orientation Ø 1 Ø 2 Average Intensity Area BPA RBF BPA RBF BPA RBF BPA RBF BPA RBF BPA RBF BPA RBF
15 INTELLIGENT FLUE GAS MONITORING IN POWER STATION BOILERS The various intelligent schemes discussed in the section 6.4 and 6.5 are implemented using MATLAB. The results obtained in the last two sections plays a vital role when it is integrated with the DCS in the real time. Hence these results are also verified using various image processing tools. This includes feature extraction using Image J. The number of features is reduced using Support Vector Machine (SVM) and Principal Component Analysis (PCA). A combination of image processing algorithms with Bayesian and intelligent classifiers are used to identify the flue gas emissions in order to ensure complete combustion. The classification of the flame images is achieved from the selected features using the intelligent and Bayesian classifiers as in Figure The flame images are collected from the control room of a boiler in the power station where forty eight correct images are identified, preprocessed and features are extracted which are reduced using SVM so as to reduce the computational complexity. Training the Bayesian, RBF and MLP classifiers have been done with 39 images taken from class 1, class 2 and class 3. For testing the classifier s performance, 9 images are considered, three from each combustion category. The comparison of various algorithms during testing, indicate that the intelligent classifier gives maximum classification performance as compared to Bayesian classifier. The SVM feature reduction with intelligent classifier yields optimal values for true positive, false positive, recall and precision. The classification performance is also validated by cross validation. The proposed algorithm is used to provide an intelligent combustion quality monitoring technique in a feed forward manner thereby preventing excess emission of flue gases.
16 150 Record the flame images, its temperature and readings of the flue gas emissions from the gas analyzers Pre-processing and Feature Extraction Training Feature Reduction Classification Testing Reduced Feature set Classification Validation Figure 6.10 Block diagram for implementation of flame monitoring in power station boilers using WEKA Procedure for Implementation of Flue Gas Monitoring from Flame Image Analysis using (Knowledge Flow (KFL) in WEKA The knowledge flow in the Weka tool is used for further analysis of the flame images, which gives a visual design of the various blocks in implementing the project. The steps involved include by adding the required nodes like Comma Separated Values (CSV) loader, Class Assigner, Cross validation fold maker, Radial Basis Function (RBF) classifier, Classifier
17 151 performance evaluator and text viewer. Secondly, connect the nodes one after the other and run the process using default steps for each node and view the results in the text viewer after executing the command Start loading Loading of the feature set as Comma Separated Values (CSV) The feature set in the excel sheet can be stored in the form of ARFF (Attribute Relation file Format), CSV, C4.5 or binary file. Using the data set and configure options, the feature file (CSV file) can be loaded and connected to the next block. The CSV loader is a part of the data sources Class Assigner and Cross Validation This block helps to assign the class to the loaded data. In this case the class refers to the three categories of combustion namely the complete, partial and incomplete combustion. Then the class assigner block is connected. Cross validation is a way to predict the fit of a model to a hypothetical validation set when an explicit validation set is not available. Using cross validation, it is possible to compare the methods in terms of their respective fractions of misclassified characters Radial Basis Function Classifier (RBF) In this intelligent classifier used is a RBF classifier. The Radial Basis function (RBF) network is a feed forward neural network with input layer, hidden layer and output layer. The centres are created from the training data by which the distance between the training pattern and the centres are found. The output of the hidden layer is obtained by using the Gaussian function. The centres should be chosen such that all the patterns in a group are around the respective centres. The network architecture is 11 x 3 x 1 for measurement of each and every output (flue gases, flame temperature,
18 152 combustion quality and air/flue ratio). The Figures 6.11 shows the various stages of building the KFL model for flue gas monitoring using RBF classifier. Figure 6.11 Various stages for RBF classifier implementation using WEKA, KFL Multilayer Perceptron (MLP) Another type of intelligent classifier called MultiLayer Perceptron (MLP) classifier has been used for inferring combustion quality in power station boilers. The MLP has three layers namely the input layer, hidden layer and the output layer. The activation function used is a sigmoid function. The number of nodes in the input layer is six, number of nodes in the hidden layer is twenty and that in the output layer are three to identify the combustion quality. The mean squared error is the objective function. Similar methods are followed as that of RBF classifier so as to obtain results for MLP classifier.
19 Procedure for Implementation of Flue Gas Monitoring from Flame Image Analysis using Weka Tool (WEKA Explorer) The procedure for implementation of the flue gas monitoring from flame images include the following stages as mentioned below Preprocess: Load, analyze, and filter data Visualize: Compare pairs of attributes and Plot matrices Classify: All algorithms seem in class (Naive Bayes, etc.) Feature selection: Forward feature subset selection, etc. Classifiers allowed in assignment include decision trees, naive Bayes and linear classifiers Repeating many experiments in Weka helps to reproduce with other classifiers and parameters (e.g., inside Weka Experimenter ) and involves less time for coding and experimenting which means that there is more time for analyzing intrinsic differences between the classifiers. In preprocessing, as the data set is loaded, use the data set Editor and apply a filter to remove attributes and instances. The prediction is a linear function of the input. In case of binary predictions, a linear classifier splits a high dimensional input space with a hyper plane (i.e., a plane in 3D, or a straight line in 2D). Many popular effective classifiers are linear like perceptron, linear SVM and logistic regression like maximum entropy and exponential model.
20 154 To visualize, load a dataset and visualize it. To examine instance information, note the discrepancy in numbering between instance information and dataset viewer. Select instances and rectangles and save the new dataset to a file To classify, load the dataset and classify it with the various types of classifiers which include MLP and RBF classifier as shown Figure 6.12 The learning process includes testing and training set. Examine the classifier output panel and visualize by right clicking the entry in the result list. Interpret the classification accuracy and confusion matrix. Test the classifier on a supplied test set and visualize the classifier errors once again by right clicking the entry in the result list. Figure 6.12 Screen shot for implementing MLP classifier
21 PROCEDURE FOR VALIDATION Validation is a process of checking that an algorithm meets the specifications and that it fulfills its intended purpose. It is a quality assurance process of establishing evidence that provides a high degree of assurance that an algorithm accomplishes its intended requirements. This often involves acceptance of fitness for purpose with the end users. The validation is carried out with the flame images collected during some other period of time. The flame images pertaining to complete, partial and incomplete combustion category for validation of the various intelligent schemes are shown in Table 6.4. Table 6.4 Sample Flame images for validation Class 1 Complete combustion Class 2 Partial combustion Class 3 Incomplete combustion One of the real time problems is the need for control engineers who understand, and are competent in, the very demanding field of computer systems as well as in the more traditional areas of engineering. But whereas the quantity and variety of information required by the engineer has grown enormously over the past half century, the period allocated to graduate training has not expanded beyond the same four or five years that is being spent in training. Computers are specialised things that one might, perhaps, study after graduating. Besides being complicated, computer technology is beguiling. It is tempting, and intellectually satisfying, to sit at a keyboard
22 156 tapping away and generating words, formulae or pictures on the screen. At the worst the system may 'crash', necessitating a reboot of the process that may, at the worst it results in the loss of much carefully constructed information. On the other hand, a computer controlling any power station plant is in command of a huge process involving explosive mixtures of gases, steam at pressures and temperatures that become instantly lethal if anything goes wrong, and massive roaring turbines driving generators that produce megawatts of power. A small mistake or lack of attention to detail in such a case can have consequences that will certainly be severe, probably very expensive and possibly tragic. A power station is a complex thing, and its construction is a frantic, long drawn out process involving many people, sometimes hundreds of them, working amid the difficulties of noise, dust and dirt, and extremes of temperature. Heavy items are craned or manhandled into position under a mess of cables and pipes, often with showers of sparks raining down from welding and cutting operations high above. An instrument lovingly installed on a pipe is all too often used as a foothold for a heavy booted rigger reaching up to install an item on another pipe. Thus using the standard packages the various intelligent schemes are implemented and analyzed. The algorithms are developed using standard packages can be integrated with the DCS for online monitoring which will be a cost effective method.
Analysis Tools and Libraries for BigData
+ Analysis Tools and Libraries for BigData Lecture 02 Abhijit Bendale + Office Hours 2 n Terry Boult (Waiting to Confirm) n Abhijit Bendale (Tue 2:45 to 4:45 pm). Best if you email me in advance, but I
More informationChapter 6. The stacking ensemble approach
82 This chapter proposes the stacking ensemble approach for combining different data mining classifiers to get better performance. Other combination techniques like voting, bagging etc are also described
More informationAnalecta Vol. 8, No. 2 ISSN 2064-7964
EXPERIMENTAL APPLICATIONS OF ARTIFICIAL NEURAL NETWORKS IN ENGINEERING PROCESSING SYSTEM S. Dadvandipour Institute of Information Engineering, University of Miskolc, Egyetemváros, 3515, Miskolc, Hungary,
More informationSocial Media Mining. Data Mining Essentials
Introduction Data production rate has been increased dramatically (Big Data) and we are able store much more data than before E.g., purchase data, social media data, mobile phone data Businesses and customers
More informationThe Scientific Data Mining Process
Chapter 4 The Scientific Data Mining Process When I use a word, Humpty Dumpty said, in rather a scornful tone, it means just what I choose it to mean neither more nor less. Lewis Carroll [87, p. 214] In
More information8. Machine Learning Applied Artificial Intelligence
8. Machine Learning Applied Artificial Intelligence Prof. Dr. Bernhard Humm Faculty of Computer Science Hochschule Darmstadt University of Applied Sciences 1 Retrospective Natural Language Processing Name
More informationUniversité de Montpellier 2 Hugo Alatrista-Salas : hugo.alatrista-salas@teledetection.fr
Université de Montpellier 2 Hugo Alatrista-Salas : hugo.alatrista-salas@teledetection.fr WEKA Gallirallus Zeland) australis : Endemic bird (New Characteristics Waikato university Weka is a collection
More information1. Classification problems
Neural and Evolutionary Computing. Lab 1: Classification problems Machine Learning test data repository Weka data mining platform Introduction Scilab 1. Classification problems The main aim of a classification
More informationWhat is Data Mining, and How is it Useful for Power Plant Optimization? (and How is it Different from DOE, CFD, Statistical Modeling)
data analysis data mining quality control web-based analytics What is Data Mining, and How is it Useful for Power Plant Optimization? (and How is it Different from DOE, CFD, Statistical Modeling) StatSoft
More informationData quality in Accounting Information Systems
Data quality in Accounting Information Systems Comparing Several Data Mining Techniques Erjon Zoto Department of Statistics and Applied Informatics Faculty of Economy, University of Tirana Tirana, Albania
More informationDATA MINING TOOL FOR INTEGRATED COMPLAINT MANAGEMENT SYSTEM WEKA 3.6.7
DATA MINING TOOL FOR INTEGRATED COMPLAINT MANAGEMENT SYSTEM WEKA 3.6.7 UNDER THE GUIDANCE Dr. N.P. DHAVALE, DGM, INFINET Department SUBMITTED TO INSTITUTE FOR DEVELOPMENT AND RESEARCH IN BANKING TECHNOLOGY
More informationProf. Pietro Ducange Students Tutor and Practical Classes Course of Business Intelligence 2014 http://www.iet.unipi.it/p.ducange/esercitazionibi/
Prof. Pietro Ducange Students Tutor and Practical Classes Course of Business Intelligence 2014 http://www.iet.unipi.it/p.ducange/esercitazionibi/ Email: p.ducange@iet.unipi.it Office: Dipartimento di Ingegneria
More informationAn Introduction to WEKA. As presented by PACE
An Introduction to WEKA As presented by PACE Download and Install WEKA Website: http://www.cs.waikato.ac.nz/~ml/weka/index.html 2 Content Intro and background Exploring WEKA Data Preparation Creating Models/
More informationAzure Machine Learning, SQL Data Mining and R
Azure Machine Learning, SQL Data Mining and R Day-by-day Agenda Prerequisites No formal prerequisites. Basic knowledge of SQL Server Data Tools, Excel and any analytical experience helps. Best of all:
More informationAn Introduction to Data Mining
An Introduction to Intel Beijing wei.heng@intel.com January 17, 2014 Outline 1 DW Overview What is Notable Application of Conference, Software and Applications Major Process in 2 Major Tasks in Detail
More informationPredict Influencers in the Social Network
Predict Influencers in the Social Network Ruishan Liu, Yang Zhao and Liuyu Zhou Email: rliu2, yzhao2, lyzhou@stanford.edu Department of Electrical Engineering, Stanford University Abstract Given two persons
More informationA New Approach For Estimating Software Effort Using RBFN Network
IJCSNS International Journal of Computer Science and Network Security, VOL.8 No.7, July 008 37 A New Approach For Estimating Software Using RBFN Network Ch. Satyananda Reddy, P. Sankara Rao, KVSVN Raju,
More informationPractical Data Science with Azure Machine Learning, SQL Data Mining, and R
Practical Data Science with Azure Machine Learning, SQL Data Mining, and R Overview This 4-day class is the first of the two data science courses taught by Rafal Lukawiecki. Some of the topics will be
More informationPredictive Dynamix Inc
Predictive Modeling Technology Predictive modeling is concerned with analyzing patterns and trends in historical and operational data in order to transform data into actionable decisions. This is accomplished
More informationFinal Project Report
CPSC545 by Introduction to Data Mining Prof. Martin Schultz & Prof. Mark Gerstein Student Name: Yu Kor Hugo Lam Student ID : 904907866 Due Date : May 7, 2007 Introduction Final Project Report Pseudogenes
More informationKeywords Data mining, Classification Algorithm, Decision tree, J48, Random forest, Random tree, LMT, WEKA 3.7. Fig.1. Data mining techniques.
International Journal of Emerging Research in Management &Technology Research Article October 2015 Comparative Study of Various Decision Tree Classification Algorithm Using WEKA Purva Sewaiwar, Kamal Kant
More informationIntroduction Predictive Analytics Tools: Weka
Introduction Predictive Analytics Tools: Weka Predictive Analytics Center of Excellence San Diego Supercomputer Center University of California, San Diego Tools Landscape Considerations Scale User Interface
More informationData Mining Algorithms Part 1. Dejan Sarka
Data Mining Algorithms Part 1 Dejan Sarka Join the conversation on Twitter: @DevWeek #DW2015 Instructor Bio Dejan Sarka (dsarka@solidq.com) 30 years of experience SQL Server MVP, MCT, 13 books 7+ courses
More informationArtificial Neural Network, Decision Tree and Statistical Techniques Applied for Designing and Developing E-mail Classifier
International Journal of Recent Technology and Engineering (IJRTE) ISSN: 2277-3878, Volume-1, Issue-6, January 2013 Artificial Neural Network, Decision Tree and Statistical Techniques Applied for Designing
More informationNeural Networks and Support Vector Machines
INF5390 - Kunstig intelligens Neural Networks and Support Vector Machines Roar Fjellheim INF5390-13 Neural Networks and SVM 1 Outline Neural networks Perceptrons Neural networks Support vector machines
More informationData Mining with Weka
Data Mining with Weka Class 1 Lesson 1 Introduction Ian H. Witten Department of Computer Science University of Waikato New Zealand weka.waikato.ac.nz Data Mining with Weka a practical course on how to
More informationMachine learning for algo trading
Machine learning for algo trading An introduction for nonmathematicians Dr. Aly Kassam Overview High level introduction to machine learning A machine learning bestiary What has all this got to do with
More informationAUTOMATION OF ENERGY DEMAND FORECASTING. Sanzad Siddique, B.S.
AUTOMATION OF ENERGY DEMAND FORECASTING by Sanzad Siddique, B.S. A Thesis submitted to the Faculty of the Graduate School, Marquette University, in Partial Fulfillment of the Requirements for the Degree
More informationA Content based Spam Filtering Using Optical Back Propagation Technique
A Content based Spam Filtering Using Optical Back Propagation Technique Sarab M. Hameed 1, Noor Alhuda J. Mohammed 2 Department of Computer Science, College of Science, University of Baghdad - Iraq ABSTRACT
More informationPATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION
PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION Introduction In the previous chapter, we explored a class of regression models having particularly simple analytical
More informationT-61.3050 : Email Classification as Spam or Ham using Naive Bayes Classifier. Santosh Tirunagari : 245577
T-61.3050 : Email Classification as Spam or Ham using Naive Bayes Classifier Santosh Tirunagari : 245577 January 20, 2011 Abstract This term project gives a solution how to classify an email as spam or
More informationData Quality Mining: Employing Classifiers for Assuring consistent Datasets
Data Quality Mining: Employing Classifiers for Assuring consistent Datasets Fabian Grüning Carl von Ossietzky Universität Oldenburg, Germany, fabian.gruening@informatik.uni-oldenburg.de Abstract: Independent
More informationAn Introduction to Data Mining. Big Data World. Related Fields and Disciplines. What is Data Mining? 2/12/2015
An Introduction to Data Mining for Wind Power Management Spring 2015 Big Data World Every minute: Google receives over 4 million search queries Facebook users share almost 2.5 million pieces of content
More informationAdvanced analytics at your hands
2.3 Advanced analytics at your hands Neural Designer is the most powerful predictive analytics software. It uses innovative neural networks techniques to provide data scientists with results in a way previously
More informationData Mining mit der JMSL Numerical Library for Java Applications
Data Mining mit der JMSL Numerical Library for Java Applications Stefan Sineux 8. Java Forum Stuttgart 07.07.2005 Agenda Visual Numerics JMSL TM Numerical Library Neuronale Netze (Hintergrund) Demos Neuronale
More informationBank Customers (Credit) Rating System Based On Expert System and ANN
Bank Customers (Credit) Rating System Based On Expert System and ANN Project Review Yingzhen Li Abstract The precise rating of customers has a decisive impact on loan business. We constructed the BP network,
More informationIs a Data Scientist the New Quant? Stuart Kozola MathWorks
Is a Data Scientist the New Quant? Stuart Kozola MathWorks 2015 The MathWorks, Inc. 1 Facts or information used usually to calculate, analyze, or plan something Information that is produced or stored by
More informationLecture 6. Artificial Neural Networks
Lecture 6 Artificial Neural Networks 1 1 Artificial Neural Networks In this note we provide an overview of the key concepts that have led to the emergence of Artificial Neural Networks as a major paradigm
More informationPredicting the Risk of Heart Attacks using Neural Network and Decision Tree
Predicting the Risk of Heart Attacks using Neural Network and Decision Tree S.Florence 1, N.G.Bhuvaneswari Amma 2, G.Annapoorani 3, K.Malathi 4 PG Scholar, Indian Institute of Information Technology, Srirangam,
More informationANN Based Fault Classifier and Fault Locator for Double Circuit Transmission Line
International Journal of Computer Sciences and Engineering Open Access Research Paper Volume-4, Special Issue-2, April 2016 E-ISSN: 2347-2693 ANN Based Fault Classifier and Fault Locator for Double Circuit
More information1 Topic. 2 Scilab. 2.1 What is Scilab?
1 Topic Data Mining with Scilab. I know the name "Scilab" for a long time (http://www.scilab.org/en). For me, it is a tool for numerical analysis. It seemed not interesting in the context of the statistical
More informationNeural network software tool development: exploring programming language options
INEB- PSI Technical Report 2006-1 Neural network software tool development: exploring programming language options Alexandra Oliveira aao@fe.up.pt Supervisor: Professor Joaquim Marques de Sá June 2006
More informationThe Role of Size Normalization on the Recognition Rate of Handwritten Numerals
The Role of Size Normalization on the Recognition Rate of Handwritten Numerals Chun Lei He, Ping Zhang, Jianxiong Dong, Ching Y. Suen, Tien D. Bui Centre for Pattern Recognition and Machine Intelligence,
More informationFace Recognition For Remote Database Backup System
Face Recognition For Remote Database Backup System Aniza Mohamed Din, Faudziah Ahmad, Mohamad Farhan Mohamad Mohsin, Ku Ruhana Ku-Mahamud, Mustafa Mufawak Theab 2 Graduate Department of Computer Science,UUM
More informationComparison of K-means and Backpropagation Data Mining Algorithms
Comparison of K-means and Backpropagation Data Mining Algorithms Nitu Mathuriya, Dr. Ashish Bansal Abstract Data mining has got more and more mature as a field of basic research in computer science and
More informationDetection. Perspective. Network Anomaly. Bhattacharyya. Jugal. A Machine Learning »C) Dhruba Kumar. Kumar KaKta. CRC Press J Taylor & Francis Croup
Network Anomaly Detection A Machine Learning Perspective Dhruba Kumar Bhattacharyya Jugal Kumar KaKta»C) CRC Press J Taylor & Francis Croup Boca Raton London New York CRC Press is an imprint of the Taylor
More informationA Partially Supervised Metric Multidimensional Scaling Algorithm for Textual Data Visualization
A Partially Supervised Metric Multidimensional Scaling Algorithm for Textual Data Visualization Ángela Blanco Universidad Pontificia de Salamanca ablancogo@upsa.es Spain Manuel Martín-Merino Universidad
More informationWEKA Explorer User Guide for Version 3-4-3
WEKA Explorer User Guide for Version 3-4-3 Richard Kirkby Eibe Frank November 9, 2004 c 2002, 2004 University of Waikato Contents 1 Launching WEKA 2 2 The WEKA Explorer 2 Section Tabs................................
More informationBOOSTING - A METHOD FOR IMPROVING THE ACCURACY OF PREDICTIVE MODEL
The Fifth International Conference on e-learning (elearning-2014), 22-23 September 2014, Belgrade, Serbia BOOSTING - A METHOD FOR IMPROVING THE ACCURACY OF PREDICTIVE MODEL SNJEŽANA MILINKOVIĆ University
More informationEFFICIENT DATA PRE-PROCESSING FOR DATA MINING
EFFICIENT DATA PRE-PROCESSING FOR DATA MINING USING NEURAL NETWORKS JothiKumar.R 1, Sivabalan.R.V 2 1 Research scholar, Noorul Islam University, Nagercoil, India Assistant Professor, Adhiparasakthi College
More informationImproving the Thermal Efficiency of Coal-Fired Power Plants: A Data Mining Approach
Paper 1805-2014 Improving the Thermal Efficiency of Coal-Fired Power Plants: A Data Mining Approach Thanrawee Phurithititanapong and Jongsawas Chongwatpol NIDA Business School, National Institute of Development
More informationLearning outcomes. Knowledge and understanding. Competence and skills
Syllabus Master s Programme in Statistics and Data Mining 120 ECTS Credits Aim The rapid growth of databases provides scientists and business people with vast new resources. This programme meets the challenges
More informationTHE COMPARISON OF DATA MINING TOOLS
T.C. İSTANBUL KÜLTÜR UNIVERSITY THE COMPARISON OF DATA MINING TOOLS Data Warehouses and Data Mining Yrd.Doç.Dr. Ayça ÇAKMAK PEHLİVANLI Department of Computer Engineering İstanbul Kültür University submitted
More informationCurrent Standard: Mathematical Concepts and Applications Shape, Space, and Measurement- Primary
Shape, Space, and Measurement- Primary A student shall apply concepts of shape, space, and measurement to solve problems involving two- and three-dimensional shapes by demonstrating an understanding of:
More informationBEHAVIOR BASED CREDIT CARD FRAUD DETECTION USING SUPPORT VECTOR MACHINES
BEHAVIOR BASED CREDIT CARD FRAUD DETECTION USING SUPPORT VECTOR MACHINES 123 CHAPTER 7 BEHAVIOR BASED CREDIT CARD FRAUD DETECTION USING SUPPORT VECTOR MACHINES 7.1 Introduction Even though using SVM presents
More informationEnglish version. Manual for Determination of Combined Heat and Power (CHP)
CEN/CENELEC WORKSHOP CWA 45547 September 2004 AGREEMENT ICS 27.100 English version Manual for Determination of Combined Heat and Power (CHP) This CEN/CENELEC Workshop Agreement has been drafted and approved
More informationKnowledge Discovery from patents using KMX Text Analytics
Knowledge Discovery from patents using KMX Text Analytics Dr. Anton Heijs anton.heijs@treparel.com Treparel Abstract In this white paper we discuss how the KMX technology of Treparel can help searchers
More informationAnalysis of WEKA Data Mining Algorithm REPTree, Simple Cart and RandomTree for Classification of Indian News
Analysis of WEKA Data Mining Algorithm REPTree, Simple Cart and RandomTree for Classification of Indian News Sushilkumar Kalmegh Associate Professor, Department of Computer Science, Sant Gadge Baba Amravati
More informationComponent Ordering in Independent Component Analysis Based on Data Power
Component Ordering in Independent Component Analysis Based on Data Power Anne Hendrikse Raymond Veldhuis University of Twente University of Twente Fac. EEMCS, Signals and Systems Group Fac. EEMCS, Signals
More informationEFFICIENCY OF DECISION TREES IN PREDICTING STUDENT S ACADEMIC PERFORMANCE
EFFICIENCY OF DECISION TREES IN PREDICTING STUDENT S ACADEMIC PERFORMANCE S. Anupama Kumar 1 and Dr. Vijayalakshmi M.N 2 1 Research Scholar, PRIST University, 1 Assistant Professor, Dept of M.C.A. 2 Associate
More informationNew Work Item for ISO 3534-5 Predictive Analytics (Initial Notes and Thoughts) Introduction
Introduction New Work Item for ISO 3534-5 Predictive Analytics (Initial Notes and Thoughts) Predictive analytics encompasses the body of statistical knowledge supporting the analysis of massive data sets.
More informationAssessment. Presenter: Yupu Zhang, Guoliang Jin, Tuo Wang Computer Vision 2008 Fall
Automatic Photo Quality Assessment Presenter: Yupu Zhang, Guoliang Jin, Tuo Wang Computer Vision 2008 Fall Estimating i the photorealism of images: Distinguishing i i paintings from photographs h Florin
More informationData Mining Techniques for Prognosis in Pancreatic Cancer
Data Mining Techniques for Prognosis in Pancreatic Cancer by Stuart Floyd A Thesis Submitted to the Faculty of the WORCESTER POLYTECHNIC INSTITUE In partial fulfillment of the requirements for the Degree
More informationLearning is a very general term denoting the way in which agents:
What is learning? Learning is a very general term denoting the way in which agents: Acquire and organize knowledge (by building, modifying and organizing internal representations of some external reality);
More informationIn this tutorial, we try to build a roc curve from a logistic regression.
Subject In this tutorial, we try to build a roc curve from a logistic regression. Regardless the software we used, even for commercial software, we have to prepare the following steps when we want build
More informationFeature Subset Selection in E-mail Spam Detection
Feature Subset Selection in E-mail Spam Detection Amir Rajabi Behjat, Universiti Technology MARA, Malaysia IT Security for the Next Generation Asia Pacific & MEA Cup, Hong Kong 14-16 March, 2012 Feature
More informationNeural Network based Vehicle Classification for Intelligent Traffic Control
Neural Network based Vehicle Classification for Intelligent Traffic Control Saeid Fazli 1, Shahram Mohammadi 2, Morteza Rahmani 3 1,2,3 Electrical Engineering Department, Zanjan University, Zanjan, IRAN
More informationInternational Journal of Computer Trends and Technology (IJCTT) volume 4 Issue 8 August 2013
A Short-Term Traffic Prediction On A Distributed Network Using Multiple Regression Equation Ms.Sharmi.S 1 Research Scholar, MS University,Thirunelvelli Dr.M.Punithavalli Director, SREC,Coimbatore. Abstract:
More informationMore Data Mining with Weka
More Data Mining with Weka Class 5 Lesson 1 Simple neural networks Ian H. Witten Department of Computer Science University of Waikato New Zealand weka.waikato.ac.nz Lesson 5.1: Simple neural networks Class
More informationBack Propagation Neural Networks User Manual
Back Propagation Neural Networks User Manual Author: Lukáš Civín Library: BP_network.dll Runnable class: NeuralNetStart Document: Back Propagation Neural Networks Page 1/28 Content: 1 INTRODUCTION TO BACK-PROPAGATION
More informationPentaho Data Mining Last Modified on January 22, 2007
Pentaho Data Mining Copyright 2007 Pentaho Corporation. Redistribution permitted. All trademarks are the property of their respective owners. For the latest information, please visit our web site at www.pentaho.org
More informationContinuous flow direct water heating for potable hot water
Continuous flow direct water heating for potable hot water An independently produced White Paper for Rinnai UK 2013 www.rinnaiuk.com In the 35 years since direct hot water systems entered the UK commercial
More informationImpelling Heart Attack Prediction System using Data Mining and Artificial Neural Network
General Article International Journal of Current Engineering and Technology E-ISSN 2277 4106, P-ISSN 2347-5161 2014 INPRESSCO, All Rights Reserved Available at http://inpressco.com/category/ijcet Impelling
More informationBachelor of Games and Virtual Worlds (Programming) Subject and Course Summaries
First Semester Development 1A On completion of this subject students will be able to apply basic programming and problem solving skills in a 3 rd generation object-oriented programming language (such as
More informationSTATISTICA. Financial Institutions. Case Study: Credit Scoring. and
Financial Institutions and STATISTICA Case Study: Credit Scoring STATISTICA Solutions for Business Intelligence, Data Mining, Quality Control, and Web-based Analytics Table of Contents INTRODUCTION: WHAT
More informationHow To Understand How Weka Works
More Data Mining with Weka Class 1 Lesson 1 Introduction Ian H. Witten Department of Computer Science University of Waikato New Zealand weka.waikato.ac.nz More Data Mining with Weka a practical course
More informationArtificial Neural Networks and Support Vector Machines. CS 486/686: Introduction to Artificial Intelligence
Artificial Neural Networks and Support Vector Machines CS 486/686: Introduction to Artificial Intelligence 1 Outline What is a Neural Network? - Perceptron learners - Multi-layer networks What is a Support
More informationEM Clustering Approach for Multi-Dimensional Analysis of Big Data Set
EM Clustering Approach for Multi-Dimensional Analysis of Big Data Set Amhmed A. Bhih School of Electrical and Electronic Engineering Princy Johnson School of Electrical and Electronic Engineering Martin
More informationContents WEKA Microsoft SQL Database
WEKA User Manual Contents WEKA Introduction 3 Background information. 3 Installation. 3 Where to get WEKA... 3 Downloading Information... 3 Opening the program.. 4 Chooser Menu. 4-6 Preprocessing... 6-7
More informationTrading Strategies and the Cat Tournament Protocol
M A C H I N E L E A R N I N G P R O J E C T F I N A L R E P O R T F A L L 2 7 C S 6 8 9 CLASSIFICATION OF TRADING STRATEGIES IN ADAPTIVE MARKETS MARK GRUMAN MANJUNATH NARAYANA Abstract In the CAT Tournament,
More informationHigh Productivity Data Processing Analytics Methods with Applications
High Productivity Data Processing Analytics Methods with Applications Dr. Ing. Morris Riedel et al. Adjunct Associate Professor School of Engineering and Natural Sciences, University of Iceland Research
More informationApplication of Neural Network in User Authentication for Smart Home System
Application of Neural Network in User Authentication for Smart Home System A. Joseph, D.B.L. Bong, D.A.A. Mat Abstract Security has been an important issue and concern in the smart home systems. Smart
More informationStructural Health Monitoring Tools (SHMTools)
Structural Health Monitoring Tools (SHMTools) Getting Started LANL/UCSD Engineering Institute LA-CC-14-046 c Copyright 2014, Los Alamos National Security, LLC All rights reserved. May 30, 2014 Contents
More informationChapter 12 Discovering New Knowledge Data Mining
Chapter 12 Discovering New Knowledge Data Mining Becerra-Fernandez, et al. -- Knowledge Management 1/e -- 2004 Prentice Hall Additional material 2007 Dekai Wu Chapter Objectives Introduce the student to
More informationProgramming Exercise 3: Multi-class Classification and Neural Networks
Programming Exercise 3: Multi-class Classification and Neural Networks Machine Learning November 4, 2011 Introduction In this exercise, you will implement one-vs-all logistic regression and neural networks
More informationGLOVE-BASED GESTURE RECOGNITION SYSTEM
CLAWAR 2012 Proceedings of the Fifteenth International Conference on Climbing and Walking Robots and the Support Technologies for Mobile Machines, Baltimore, MD, USA, 23 26 July 2012 747 GLOVE-BASED GESTURE
More informationContent-Based Recommendation
Content-Based Recommendation Content-based? Item descriptions to identify items that are of particular interest to the user Example Example Comparing with Noncontent based Items User-based CF Searches
More informationEmployer Health Insurance Premium Prediction Elliott Lui
Employer Health Insurance Premium Prediction Elliott Lui 1 Introduction The US spends 15.2% of its GDP on health care, more than any other country, and the cost of health insurance is rising faster than
More informationMaschinelles Lernen mit MATLAB
Maschinelles Lernen mit MATLAB Jérémy Huard Applikationsingenieur The MathWorks GmbH 2015 The MathWorks, Inc. 1 Machine Learning is Everywhere Image Recognition Speech Recognition Stock Prediction Medical
More informationfrom Larson Text By Susan Miertschin
Decision Tree Data Mining Example from Larson Text By Susan Miertschin 1 Problem The Maximum Miniatures Marketing Department wants to do a targeted mailing gpromoting the Mythic World line of figurines.
More informationComparison of Non-linear Dimensionality Reduction Techniques for Classification with Gene Expression Microarray Data
CMPE 59H Comparison of Non-linear Dimensionality Reduction Techniques for Classification with Gene Expression Microarray Data Term Project Report Fatma Güney, Kübra Kalkan 1/15/2013 Keywords: Non-linear
More informationActive Learning SVM for Blogs recommendation
Active Learning SVM for Blogs recommendation Xin Guan Computer Science, George Mason University Ⅰ.Introduction In the DH Now website, they try to review a big amount of blogs and articles and find the
More informationON INTEGRATING UNSUPERVISED AND SUPERVISED CLASSIFICATION FOR CREDIT RISK EVALUATION
ISSN 9 X INFORMATION TECHNOLOGY AND CONTROL, 00, Vol., No.A ON INTEGRATING UNSUPERVISED AND SUPERVISED CLASSIFICATION FOR CREDIT RISK EVALUATION Danuta Zakrzewska Institute of Computer Science, Technical
More informationBIDM Project. Predicting the contract type for IT/ITES outsourcing contracts
BIDM Project Predicting the contract type for IT/ITES outsourcing contracts N a n d i n i G o v i n d a r a j a n ( 6 1 2 1 0 5 5 6 ) The authors believe that data modelling can be used to predict if an
More informationSTA 4273H: Statistical Machine Learning
STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Statistics! rsalakhu@utstat.toronto.edu! http://www.cs.toronto.edu/~rsalakhu/ Lecture 6 Three Approaches to Classification Construct
More informationStatistical Validation and Data Analytics in ediscovery. Jesse Kornblum
Statistical Validation and Data Analytics in ediscovery Jesse Kornblum Administrivia Silence your mobile Interactive talk Please ask questions 2 Outline Introduction Big Questions What Makes Things Similar?
More informationA Hybrid Modeling Platform to meet Basel II Requirements in Banking Jeffery Morrision, SunTrust Bank, Inc.
A Hybrid Modeling Platform to meet Basel II Requirements in Banking Jeffery Morrision, SunTrust Bank, Inc. Introduction: The Basel Capital Accord, ready for implementation in force around 2006, sets out
More informationAnalytics on Big Data
Analytics on Big Data Riccardo Torlone Università Roma Tre Credits: Mohamed Eltabakh (WPI) Analytics The discovery and communication of meaningful patterns in data (Wikipedia) It relies on data analysis
More informationEvaluation of Feature Selection Methods for Predictive Modeling Using Neural Networks in Credits Scoring
714 Evaluation of Feature election Methods for Predictive Modeling Using Neural Networks in Credits coring Raghavendra B. K. Dr. M.G.R. Educational and Research Institute, Chennai-95 Email: raghavendra_bk@rediffmail.com
More informationCOMPARISON OF OBJECT BASED AND PIXEL BASED CLASSIFICATION OF HIGH RESOLUTION SATELLITE IMAGES USING ARTIFICIAL NEURAL NETWORKS
COMPARISON OF OBJECT BASED AND PIXEL BASED CLASSIFICATION OF HIGH RESOLUTION SATELLITE IMAGES USING ARTIFICIAL NEURAL NETWORKS B.K. Mohan and S. N. Ladha Centre for Studies in Resources Engineering IIT
More information