DOCSLIB.ORG

Decision Tree Rule Reduction Using Linear Classifiers in Multilayer Perceptron

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Decision Tree Rule Reduction Using Linear Classifiers in Multilayer Perceptron From: FLAIRS-01 Proceedings. Copyright © 2001, AAAI (www.aaai.org). All rights reserved. Decision Tree Rule Reduction Using Linear Classifiers in Multilayer Perceptron DaeEun Kim Sea Woo Kim Division of Informatics Dept. of Information and University of Edinburgh Communication Engineering, KAIST 5 Forrest Hill Cheongryang l-dong Edinburgh, EHI 2QL, United Kingdom Seoul, 130-011, Korea [email protected] [email protected] Abstract relationship betweendata points, using a statistical tech- nique. It generates manydata points on the response sur- It has beenshown that a neuralnetwork is better thana direct applicationof inductiontrees in modelingcom- face of the fitted curve, and then induces rules with a plex relations of input attributes in sampledata. We decision tree. This method was introduced as an alter- proposethat conciserules be extractedto supportdata native measure regarding the problem of direct applica- with input variablerelations over continuous-valuedat- tion of the induction tree to raw data (Irani & Qian 1990; tributes. Thoserelations as a set of linear classifiers Kim1991). However,it still has the problem of requiring can be obtained from neural networkmodeling based manyinduction rules to reflect the response surface. on back-propagation.Alinear classifier is derivedfrom In this paper we use a hybrid technique to combineneu- a linear combinationof input attributes and neuron ral networksand decision trees for data classification (Kim weightsin the first hiddenlayer of neural networks.It is shownin this paperthat whenwe use a decisiontree & Lee 2000). It has been shownthat neural networks are over linear classifiers extractedfrom a multilayerper- better than direct application of induction trees in mod- ceptron, the numberof rules can be reduced.We have eling nonlinear characteristics of sample data (Dietterich, tested this methodover several data sets to compareit Hild, & Bakiri 1990; Quinlan 1994; Setiono & Lie 1996; withdecision tree results. Fisher & McKusick 1989; Shavlik, Mooney, & Towell 1991). Neural networks have the advantage of being able to deal with noisy, inconsistent and incompletedata. A method Introduction to extract symbolicrules from neural networkshas been pro- The discovery of decision rules and recognition of patterns posed to increase the performanceof the decision process from data examples is one of the most challenging prob- (Andrews, Diederich, & Tickle 1996; Taha & Ghosh1999; lems in machine learning. If data points contain numeri- Fu 1994; Towcll & Shavlik Oct 1993; Setiono & Lie 1996). cal attributes,induction tree methods need the continuous- The KTalgorithm developed by Fu (Fu 1991) extracts rules valuedattributes to be madediscrete with threshold values. from subsets of connected weights with high activation in a Inductiontree algorithms such as C4.5build decision trees trained network. The Mof N algorithm clusters weights of by recursivelypartitioning theinput attribute space (Quin- the trained networkand removesinsignificant clusters with lan1996). The tree traversal from the root node to eachleaf low active weights. Then the rules are extracted from the leadsto oneconjunctive rule. Each internal node in thedeci- weights (Towell & Shavlik Oct 1993). siontree has a splittingcriterion orthreshold forcontinuous- A simple rule extraction algorithm that uses discrete acti- valuedattributes to partition some part of theinput space, vations over continuous hidden units is presented in by Se- andeach leaf represents a class related to theconditions of tiono and Taha (Setiono & Lie 1996; Taha & Ghosh1999). eachinternal node. They used in sequence a weight-decay back-propagation Approachesbased on decisiontrees involve making the over a three-layer feed-forward network, a pruning process continuous-valuedattributes discrete in inputspace, creat- to removeirrelevant connectionweights, a clustering of hid- ingmany rectangular divisions. As a result,they may have den unit activations, and extraction of rules from discrete theinability to detect data trends or desirableclassifica- unit activations. They derived symbolic rules from neural tionsurfaces. Even in thecase of multivariatemethods of networks that include oblique decision hyperplanes instead discretionwhich search in parallelfor threshold values for of general input attribute relations (Setiono & Liu 1997). morethan one continuous attribute (Fayyad & Irani1993; Also the direct conversion from neural networks to rules Kweldo& Kretowski1999), the decision rules may not re- has an exponential complexity whenusing search-based al- flectdata trends or thedecision tree may build many rules gorithm over incoming weights for each unit (Fu 1994; withthe support of a smallnumber of examplesor ignore Towell & Shavlik Oct 1993). Most of the rule extraction somedata points by dismissingthem as noisy. algorithms are used to derive rules from neuron weights A possible process is suggested to grasp the trend of the and neuronactivations in the hidden layer as a search-based data. It first tries to fit it with a given data set for the method. An instance-based rule extraction method is sug- Copyright©2001, AAAI, All rightsreserved. 48O FLAIRS-2001 gested to reduce computation time by escaping search-based Therules do not catch data clustering completelyin this ex- methods(Kim & Lee 2000). After training two hidden layer ample. Figure l(b)-(c) showsneural networkfitting with neural networks, the first hidden layer weight parametersare back-propagation method¯In Figure l(b)-(c) neural network treated as linear classifiers. Theselinear differentiated fimc- nodes have slopes alpha = 1.5, 4.0 for sigmoids, respec- tions are chosen by decision tree methodsto determine deci- tively. After curve fitting, 900 points were generated uni- sion boundariesafter re-organizing the training set in terms formly on the response surface for the mappingfrom input of the newlinear classifier attributes¯ space to class, and the response values of the neural net- Our approach is to train a neural network with sigmoid workwere calculated as shownin Figure l(d). The result functions and to use decision classifiers based on weight C4.5 to those 900 points followed the classification curves, parameters of neural networks. Then an induction tree se- but produced55 rules¯ The production of manyrules results lects the desirable input variable relations for data classifi- from the fact that decision tree makespiecewise rectangu- cation. Decision tree applications have the ability to deter- lar divisions for each rule. This happensin spite of the fact mine proper subintervals over continuousattributes by a dis- that the response surface for data clustering has a correlation cretion process. This discretion process will cover oblique betweenthe input variables. hyperplanes mentionedin Setiono’s papers. In this paper, As shownabove, the decision tree has a problemof over- we have tested linear classifiers with variable thresholds and generalization for a small number of data and an over- fixed thresholds. The methodsare tested on various types of specialization problem for a large numberof data. A pos- data and compared with the method based on the decision sible suggestion is to consider or derive relations between tree alone. input variablt s as another attribute for rule extraction¯ How- ever, it is difficult to find input variablerelations for classi- Problem Statement fication directly in supervised learning, while unsupervised Induction trees are useful for a large numberof examples, methodscan use statistical methodssuch as principal com- and they enable us to obtain proper rules from examples ponent analysis (Haykin 1999). rapidly (Quinlan 1996). However,they have the difficulty in inferring relations betweendata points and cannot handle Method noisy data. The goal for our approachis to generate rules following the shape and characteristics of response surfaces. Usually in- duction trees cannot trace the trend of data, and they deter- mine data clustering only in terms of input variables, unless we apply other relation factors or attributes. In order to im- .... f .............................. "._~.. prove classification rules from a large training data set, we allow input variable relations for multi-attributes in a set of 1 ° i rules¯ NeuralNetwork and Linear Classifiers (a) (b) We use a two-phase method for rule extraction over continuous-valuedattributes. Given a large training set of data points, the first phase, as a feature extraction phase, is to train feed-forward neural networks with back-propagation and collect the weight set over input variables in the first hidden layer. A feature useful in inferring multi-attribute relations of data is foundin the first hiddenlayer of neural networks. The extracted rules involving networkweight val- (c) ues will reflect features of data examplesand provide good classification boundaries. Also they maybe more compact Figure 1: Example(a) data set and decision boundary (O and comprehensible,compared to induction tree rules. class 1, X : class 0) (b)-(c) neural networkfitting (d) data In the second phase, as a feature combinationphase, each with 900 points (Kim & Lee 2000) extracted feature for a linear classification boundaryis com- bined together using Booleanlogic gates. In this paper, we Wecan see a simple exampleof undesirable rule extrac- use an induction tree to combineeach linear classifier. tion discoveredin the induction tree application¯ Figure1 (a) The highly nonlinear property of neural networks makes
Load more

Recommended publications
  • Warm Start for Parameter Selection of Linear Classifiers
    Warm Start for Parameter Selection of Linear Classifiers Bo-Yu Chu Chia-Hua Ho Cheng-Hao Tsai Dept. of Computer Science Dept. of Computer Science Dept. of Computer Science National Taiwan Univ., Taiwan National Taiwan Univ., Taiwan National Taiwan Univ., Taiwan [email protected] [email protected] [email protected] Chieh-Yen Lin Chih-Jen Lin Dept. of Computer Science Dept. of Computer Science National Taiwan Univ., Taiwan National Taiwan Univ., Taiwan [email protected] [email protected] ABSTRACT we may need to solve many optimization problems. Sec- In linear classification, a regularization term effectively reme- ondly, if we do not know the reasonable range of the pa- dies the overfitting problem, but selecting a good regulariza- rameters, we may need a long time to solve optimization tion parameter is usually time consuming. We consider cross problems under extreme parameter values. validation for the selection process, so several optimization In this paper, we consider using warm start to efficiently problems under different parameters must be solved. Our solve a sequence of optimization problems with different reg- aim is to devise effective warm-start strategies to efficiently ularization parameters. Warm start is a technique to reduce solve this sequence of optimization problems. We detailedly the running time of iterative methods by using the solution investigate the relationship between optimal solutions of lo- of a slightly different optimization problem as an initial point gistic regression/linear SVM and regularization parameters. for the current problem. If the initial point is close to the op- Based on the analysis, we develop an efficient tool to auto- timum, warm start is very useful.
    [Show full text]
  • Neural Networks and Backpropagation
    CS 179: LECTURE 14 NEURAL NETWORKS AND BACKPROPAGATION LAST TIME Intro to machine learning Linear regression https://en.wikipedia.org/wiki/Linear_regression Gradient descent https://en.wikipedia.org/wiki/Gradient_descent (Linear classification = minimize cross-entropy) https://en.wikipedia.org/wiki/Cross_entropy TODAY Derivation of gradient descent for linear classifier https://en.wikipedia.org/wiki/Linear_classifier Using linear classifiers to build up neural networks Gradient descent for neural networks (Back Propagation) https://en.wikipedia.org/wiki/Backpropagation REFRESHER ON THE TASK Note “Grandmother Cell” representation for {x,y} pairs. See https://en.wikipedia.org/wiki/Grandmother_cell REFRESHER ON THE TASK Find i for zi: “Best-index” -- estimated “Grandmother Cell” Neuron Can use parallel GPU reduction to find “i” for largest value. LINEAR CLASSIFIER GRADIENT We will be going through some extra steps to derive the gradient of the linear classifier -- We’ll be using the “Softmax function” https://en.wikipedia.org/wiki/Softmax_function Similarities will be seen when we start talking about neural networks LINEAR CLASSIFIER J & GRADIENT LINEAR CLASSIFIER GRADIENT LINEAR CLASSIFIER GRADIENT LINEAR CLASSIFIER GRADIENT GRADIENT DESCENT GRADIENT DESCENT, REVIEW GRADIENT DESCENT IN ND GRADIENT DESCENT STOCHASTIC GRADIENT DESCENT STOCHASTIC GRADIENT DESCENT STOCHASTIC GRADIENT DESCENT, FOR W LIMITATIONS OF LINEAR MODELS Most real-world data is not separable by a linear decision boundary Simplest example: XOR gate What if we could combine the results of multiple linear classifiers? Combine two OR gates with an AND gate to get a XOR gate ANOTHER VIEW OF LINEAR MODELS NEURAL NETWORKS NEURAL NETWORKS EXAMPLES OF ACTIVATION FNS Note that most derivatives of tanh function will be zero! Makes for much needless computation in gradient descent! MORE ACTIVATION FUNCTIONS https://medium.com/@shrutijadon10104776/survey-on- activation-functions-for-deep-learning-9689331ba092 Tanh and sigmoid used historically.
    [Show full text]
  • Lecture 2: Linear Classifiers
    Lecture 2: Linear Classifiers Andr´eMartins Deep Structured Learning Course, Fall 2018 Andr´eMartins (IST) Lecture 2 IST, Fall 2018 1 / 117 Course Information • Instructor: Andr´eMartins ([email protected]) • TAs/Guest Lecturers: Erick Fonseca & Vlad Niculae • Location: LT2 (North Tower, 4th floor) • Schedule: Wednesdays 14:30{18:00 • Communication: piazza.com/tecnico.ulisboa.pt/fall2018/pdeecdsl Andr´eMartins (IST) Lecture 2 IST, Fall 2018 2 / 117 Announcements Homework 1 is out! • Deadline: October 10 (two weeks from now) • Start early!!! List of potential projects will be sent out soon! • Deadline for project proposal: October 17 (three weeks from now) • Teams of 3 people Andr´eMartins (IST) Lecture 2 IST, Fall 2018 3 / 117 Today's Roadmap Before talking about deep learning, let us talk about shallow learning: • Supervised learning: binary and multi-class classification • Feature-based linear classifiers • Rosenblatt's perceptron algorithm • Linear separability and separation margin: perceptron's mistake bound • Other linear classifiers: naive Bayes, logistic regression, SVMs • Regularization and optimization • Limitations of linear classifiers: the XOR problem • Kernel trick. Gaussian and polynomial kernels. Andr´eMartins (IST) Lecture 2 IST, Fall 2018 4 / 117 Fake News Detection Task: tell if a news article / quote is fake or real. This is a binary classification problem. Andr´eMartins (IST) Lecture 2 IST, Fall 2018 5 / 117 Fake Or Real? Andr´eMartins (IST) Lecture 2 IST, Fall 2018 6 / 117 Fake Or Real? Andr´eMartins (IST) Lecture 2 IST, Fall 2018 7 / 117 Fake Or Real? Andr´eMartins (IST) Lecture 2 IST, Fall 2018 8 / 117 Fake Or Real? Andr´eMartins (IST) Lecture 2 IST, Fall 2018 9 / 117 Fake Or Real? Can a machine determine this automatically? Can be a very hard problem, since fact-checking is hard and requires combining several knowledge sources ..
    [Show full text]
  • Hyperplane Based Classification: Perceptron and (Intro
    Hyperplane based Classification: Perceptron and (Intro to) Support Vector Machines Piyush Rai CS5350/6350: Machine Learning September 8, 2011 (CS5350/6350) Hyperplane based Classification September8,2011 1/20 Hyperplane Separates a D-dimensional space into two half-spaces Defined by an outward pointing normal vector w RD ∈ (CS5350/6350) Hyperplane based Classification September8,2011 2/20 Hyperplane Separates a D-dimensional space into two half-spaces Defined by an outward pointing normal vector w RD ∈ w is orthogonal to any vector lying on the hyperplane (CS5350/6350) Hyperplane based Classification September8,2011 2/20 Hyperplane Separates a D-dimensional space into two half-spaces Defined by an outward pointing normal vector w RD ∈ w is orthogonal to any vector lying on the hyperplane Assumption: The hyperplane passes through origin. (CS5350/6350) Hyperplane based Classification September8,2011 2/20 Hyperplane Separates a D-dimensional space into two half-spaces Defined by an outward pointing normal vector w RD ∈ w is orthogonal to any vector lying on the hyperplane Assumption: The hyperplane passes through origin. If not, have a bias term b; we will then need both w and b to define it b > 0 means moving it parallely along w (b < 0 means in opposite direction) (CS5350/6350) Hyperplane based Classification September8,2011 2/20 Linear Classification via Hyperplanes Linear Classifiers: Represent the decision boundary by a hyperplane w For binary classification, w is assumed to point towards the positive class (CS5350/6350) Hyperplane based Classification
    [Show full text]
  • 2 Linear Classifiers and Perceptrons
    Linear Classifiers and Perceptrons 7 2 Linear Classifiers and Perceptrons CLASSIFIERS You are given sample of n observations, each with d features [aka predictors]. Some observations belong to class C; some do not. Example: Observations are bank loans Features are income & age (d = 2) Some are in class “defaulted,” some are not Goal: Predict whether future borrowers will default, based on their income & age. Represent each observation as a point in d-dimensional space, called a sample point / a feature vector / independent variables. overfitting X X X X X X X C X X X X X C X C C X C C X C X C C C income income income X C C X C X C X X X C C X C C C X C C C C C age age age [Draw this by hand; decision boundaries last. classify3.pdf ] [We draw these lines/curves separating C’s from X’s. Then we use these curves to predict which future borrowers will default. In the last example, though, we’re probably overfitting, which could hurt our predic- tions.] decision boundary: the boundary chosen by our classifier to separate items in the class from those not. overfitting: When sinuous decision boundary fits sample points so well that it doesn’t classify future points well. [A reminder that underlined phrases are definitions, worth memorizing.] Some (not all) classifiers work by computing a decision function: A function f (x) that maps a point x to a scalar such that f (x) > 0 if x class C; 2 f (x) 0 if x < class C.
    [Show full text]
  • Machine Learning Empirical Risk Minimization
    Machine Learning Empirical Risk Minimization Dmitrij Schlesinger WS2014/2015, 08.12.2014 Recap – tasks considered before n For a training dataset L = (xl, yl) ... with xl ∈ R (data) and yl ∈ {+1, −1} (classes) fing a separating hyperplane yl · [hw, xli + b] ≥ 0 ∀l → Perceptron algorithm The goal is to fing a "corridor" (stripe) of the maximal width that separates the data → Large Margin learning, linear SVM 1 kwk2 → min 2 w s.t. yl[hw, xli + b] ≥ 1 In both cases the data is assumed to be separable. What if not? ML: Empirical Risk Minimization 08.12.2014 2 Empirical Risk Minimization Let a loss function C(y, y0) be given that penalizes deviations between the true class and the estimated one (like the loss in the Bayesian Decision theory). The Empirical Risk of a decision strategy is the total loss over the training set: X R(e) = C yl, e(xl) → min e l It should bi minimized with respect to the decision strategy e. Special case (today): – the set of decisions is {+1, −1} – the loss is the delta-function C(y, y0) = δ(y 6= y0) – the decision strategy can be expressed in the form e(x) = sign f(x) with an evaluation function f : X → R Example: f(x) = hw, xi − b is a linear classifier. ML: Empirical Risk Minimization 08.12.2014 3 Hinge Loss The problem: the subject is not convex The way out: replace the real loss by its convex upper bound ← example for y = 1 (for y = −1 it should be flipped) δ y 6= sign f(x) ≤ max 0, 1 − y · f(x) It is called Hinge Loss ML: Empirical Risk Minimization 08.12.2014 4 Sub-gradient algorithm Let the evaluation function be parameterized, i.e.
    [Show full text]
  • Using Linear Classifier Probes
    Under review as a conference paper at ICLR 2017 UNDERSTANDING INTERMEDIATE LAYERS USING LINEAR CLASSIFIER PROBES Guillaume Alain & Yoshua Bengio Department of Computer Science and Operations Research Universite´ de Montreal´ Montreal, QC. H3C 3J7 [email protected] ABSTRACT Neural network models have a reputation for being black boxes. We propose a new method to better understand the roles and dynamics of the intermediate layers. This has direct consequences on the design of such models and it enables the expert to be able to justify certain heuristics (such as adding auxiliary losses in middle layers). Our method uses linear classifiers, referred to as “probes”, where a probe can only use the hidden units of a given intermediate layer as discriminating features. Moreover, these probes cannot affect the training phase of a model, and they are generally added after training. They allow the user to visualize the state of the model at multiple steps of training. We demonstrate how this can be used to develop a better intuition about models and to diagnose potential problems. 1 INTRODUCTION The recent history of deep neural networks features an impressive number of new methods and technological improvements to allow the training of deeper and more powerful networks. Despite this, models still have a reputation for being black boxes. Neural networks are criticized for their lack of interpretability, which is a tradeoff that we accept because of their amazing performance on many tasks. Efforts have been made to identify the role played by each layer, but it can be hard to find a meaning to individual layers.
    [Show full text]
  • Chapter 2 Classification
    Chapter 2 Classification In this chapter we study one of the most basic topics in machine learning called classi- fication. By classification we meant a supervised learning procedure where we try to classify a sample based on a model learned from the data in a training dataset. The goal of this chapter is to understand the general principle of classification, to explore various methodologies so that we can do classification, and to make connections between different methods. 2.1 Discriminant Function Basic Terminologies A classification method always starts with a pair of variables (x; y). The first variable x 2 X ⊂ Rd is an input vector. The set X is called the input space. Depending on the application, an input could the intensity of a pixel, a wavelet coefficient, the phase angle of wave, or anything we want to model. We assume that each input vector has a dimensionality d and d is finite. The second variable y 2 f1; : : : ; kg = Y denotes the class label of the classes fC1;:::; Ckg. The choice of the class labels is arbitrary. They do not need to be positive integers. For example, in binary classification, one can choose y 2 f0; 1g or y 2 {−1; +1g, depending which one is more convenient mathematically. In supervised learning, we assume that we have a training set D. A training set is a collection of paired variables (xj; yj), for j = 1; : : : ; n. The vector xj denotes the j-th sample input in D, and yj denotes the corresponding class label. The relationship between xj and yj is specified by the target function f : X!Y such that yj = f(xj).
    [Show full text]
  • Linear Classification Overview Logistic Regression Model I Logistic
    Overview I Logistic regression model I Linear classifiers I Gradient descent and SGD Linear classification I Multinomial logistic regression model COMS 4771 Fall 2019 0 / 27 1 / 27 Logistic regression model I Logistic regression model II d t I Suppose x is given by d real-valued features, so x R , while I Sigmoid function σ(t) := 1/(1 + e− ) ∈ y 1, +1 . I Useful property: 1 σ(t) = σ( t) ∈ {− } − − I Logistic regression model for (X,Y ): I Pr(Y = +1 X = x) = σ(xTw) | T T I Y X = x is Bernoulli (but taking values in 1, +1 rather than I Pr(Y = 1 X = x) = 1 σ(x w) = σ( x w) | {− } − | − − 0, 1 ), with “heads probability” parameter I Convenient formula: for each y 1, +1 , { } ∈ {− } 1 . Pr(Y = y X = x) = σ(yxTw). 1 + exp( xTw) | − d I w R is parameter vector of interest 1 ∈ I w not involved in marginal distribution of X (which we don’t care 0.8 much about) 0.6 0.4 0.2 0 -6 -4 -2 0 2 4 6 Figure 1: Sigmoid function 2 / 27 3 / 27 Log-odds in logistic regression model Optimal classifier in logistic regression model I Log-odds in the model is given by a linear function: I Recall that Bayes classifier is Pr(Y = +1 X = x) T ln | = x w. ? +1 if Pr(Y = +1 X = x) > 1/2 Pr(Y = 1 X = x) f (x) = | − | 1 otherwise. (− I Just like in linear regression, common to use feature expansion! d+1 If distribution of (X,Y ) comes from logistic regression model with I E.g., affine feature expansion ϕ(x) = (1, x) R I ∈ parameter w, then Bayes classifier is +1 if xTw > 0 f ?(x) = 1 otherwise.
    [Show full text]
  • Differentially Private Empirical Risk Minimization
    JournalofMachineLearningResearch12(2011)1069-1109 Submitted 6/10; Revised 2/11; Published 3/11 Differentially Private Empirical Risk Minimization Kamalika Chaudhuri [email protected] Department of Computer Science and Engineering University of California, San Diego La Jolla, CA 92093, USA Claire Monteleoni [email protected] Center for Computational Learning Systems Columbia University New York, NY 10115, USA Anand D. Sarwate [email protected] Information Theory and Applications Center University of California, San Diego La Jolla, CA 92093-0447, USA Editor: Nicolas Vayatis Abstract Privacy-preserving machine learning algorithms are crucial for the increasingly common setting in which personal data, such as medical or financial records, are analyzed. We provide general techniques to produce privacy-preserving approximations of classifiers learned via (regularized) empirical risk minimization (ERM). These algorithms are private under the ε-differential privacy definition due to Dwork et al. (2006). First we apply the output perturbation ideas of Dwork et al. (2006), to ERM classification. Then we propose a new method, objective perturbation, for privacy-preserving machine learning algorithm design. This method entails perturbing the objective function before optimizing over classifiers. If the loss and regularizer satisfy certain convexity and differentiability criteria, we prove theoretical results showing that our algorithms preserve privacy, and provide generalization bounds for linear and nonlinear kernels. We further present a privacy- preserving technique for tuning the parameters in general machine learning algorithms, thereby providing end-to-end privacy guarantees for the training process. We apply these results to produce privacy-preserving analogues of regularized logistic regression and support vector machines. We obtain encouraging results from evaluating their performance on real demographic and benchmark data sets.
    [Show full text]
  • Tutorial on Support Vector Machine (SVM) Vikramaditya Jakkula, School of EECS, Washington State University, Pullman 99164
    Tutorial on Support Vector Machine (SVM) Vikramaditya Jakkula, School of EECS, Washington State University, Pullman 99164. Abstract: In this tutorial we present a brief introduction to SVM, and we discuss about SVM from published papers, workshop materials & material collected from books and material available online on the World Wide Web. In the beginning we try to define SVM and try to talk as why SVM, with a brief overview of statistical learning theory. The mathematical formulation of SVM is presented, and theory for the implementation of SVM is briefly discussed. Finally some conclusions on SVM and application areas are included. Support Vector Machines (SVMs) are competing with Neural Networks as tools for solving pattern recognition problems. This tutorial assumes you are familiar with concepts of Linear Algebra, real analysis and also understand the working of neural networks and have some background in AI. Introduction Machine Learning is considered as a subfield of Artificial Intelligence and it is concerned with the development of techniques and methods which enable the computer to learn. In simple terms development of algorithms which enable the machine to learn and perform tasks and activities. Machine learning overlaps with statistics in many ways. Over the period of time many techniques and methodologies were developed for machine learning tasks [1]. Support Vector Machine (SVM) was first heard in 1992, introduced by Boser, Guyon, and Vapnik in COLT-92. Support vector machines (SVMs) are a set of related supervised learning methods used for classification and regression [1]. They belong to a family of generalized linear classifiers. In another terms, Support Vector Machine (SVM) is a classification and regression prediction tool that uses machine learning theory to maximize predictive accuracy while automatically avoiding over-fit to the data.
    [Show full text]
  • 2DI70 - Statistical Learning Theory Lecture Notes
    2DI70 - Statistical Learning Theory Lecture Notes Rui Castro April 3, 2018 Some of the material in these notes will be published by Cambridge University Press as Statistical Machine Learning: A Gentle Primer by Rui M. Castro and Robert D. Nowak. This early draft is free to view and download for personal use only. Not for re-distribution, re-sale or use in derivative works. c Rui M. Castro and Robert D. Nowak, 2017. 2 Contents Contents 2 1 Introduction 9 1.1 Learning from Data..................................9 1.1.1 Data Spaces...................................9 1.1.2 Loss Functions................................. 10 1.1.3 Probability Measure and Expectation.................... 11 1.1.4 Statistical Risk................................. 11 1.1.5 The Learning Problem............................. 12 2 Binary Classification and Regression 15 2.1 Binary Classification.................................. 15 2.2 Regression........................................ 18 2.3 Empirical Risk Minimization............................. 19 2.4 Model Complexity and Overfitting.......................... 21 2.5 Exercises........................................ 25 3 Competing Goals: approximation vs. estimation 29 3.1 Strategies To Avoid Overfitting............................ 30 3.1.1 Method of Sieves................................ 31 3.1.2 Complexity Penalization Methods...................... 32 3.1.3 Hold-out Methods............................... 33 4 Estimation of Lipschitz smooth functions 37 4.1 Setting.......................................... 37 4.2 Analysis........................................
    [Show full text]