Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Efficient Optimal Linear Boosting of a Pair of Classifiers.

Slides:



Advertisements
Similar presentations
國立雲林科技大學 National Yunlin University of Science and Technology Application of LVQ to novelty detection using outlier training data Hyoung-joo Lee, Sungzoon.
Advertisements

Intelligent Database Systems Lab Advisor : Dr.Hsu Graduate : Keng-Wei Chang Author : Gianfranco Chicco, Roberto Napoli Federico Piglione, Petru Postolache.
國立雲林科技大學 National Yunlin University of Science and Technology Predicting adequacy of vancomycin regimens: A learning-based classification approach to improving.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Presenter : Yu Cheng Chen Author: Hichem.
Intelligent Database Systems Lab N.Y.U.S.T. I. M. Fast exact k nearest neighbors search using an orthogonal search tree Presenter : Chun-Ping Wu Authors.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Text classification based on multi-word with support vector.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Unsupervised pattern recognition models for mixed feature-type.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Human eye sclera detection and tracking using a modified.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 On-line Learning of Sequence Data Based on Self-Organizing.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Graph self-organizing maps for cyclic and unbounded graphs.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Detecting, Assessing and Monitoring Relevant Topics in Virtual.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Data mining for credit card fraud: A comparative study.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Adaptive nonlinear manifolds and their applications to pattern.
Intelligent Database Systems Lab Advisor : Dr. Hsu Graduate : Chien-Shing Chen Author : Satoshi Oyama Takashi Kokubo Toru lshida 國立雲林科技大學 National Yunlin.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 A Comparison of SOM Based Document Categorization Systems.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology HE-Tree: a framework for detecting changes in clustering.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 The k-means range algorithm for personalized data clustering.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology A Taxonomy of Similarity Mechanisms for Case-Based Reasoning.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Mining Positive and Negative Patterns for Relevance Feature.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Looking inside self-organizing map ensembles with resampling.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology CONTOUR: an efficient algorithm for discovering discriminating.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology On Data Labeling for Clustering Categorical Data Hung-Leng.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Extracting meaningful labels for WEBSOM text archives Advisor.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Topology Preservation in Self-Organizing Feature Maps: Exact.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology A self-organizing neural network using ideas from the immune.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Presenter : Keng-Wei Chang Author: Yehuda.
Intelligent Database Systems Lab N.Y.U.S.T. I. M. An IPC-based vector space model for patent retrieval Presenter: Jun-Yi Wu Authors: Yen-Liang Chen, Yu-Ting.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 AC-ViSOM: Hybridising the Modified Adaptive Coordinate.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Exploiting Data Topology in Visualization and Clustering.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 GMDH-based feature ranking and selection for improved.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Graduate : Yu Cheng Chen Author: Manoranjan.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 The Evolving Tree — Analysis and Applications Advisor.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 A Study on Automatic Recognition of Road Signs Presenter.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 2007.SIGIR.8 New Event Detection Based on Indexing-tree.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Fast accurate fuzzy clustering through data reduction Advisor.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Utilizing Marginal Net Utility for Recommendation in E-commerce.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Graduate : Yu Cheng Chen Author: Chung-hung.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Fuzzy integration of structure adaptive SOMs for web content.
Intelligent Database Systems Lab N.Y.U.S.T. I. M. The application of SOM as a decision support tool to identify AACSB peer schools Presenter : Chun-Ping.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Graduate : Sheng-Hsuan Wang Authors :
Intelligent Database Systems Lab Advisor : Dr.Hsu Graduate : Keng-Wei Chang Author : Lian Yan and David J. Miller 國立雲林科技大學 National Yunlin University of.
Intelligent Database Systems Lab Advisor : Dr. Hsu Graduate : Chien-Shing Chen Author : Juan D.Velasquez Richard Weber Hiroshi Yasuda 國立雲林科技大學 National.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Rival-Model Penalized Self-Organizing Map Yiu-ming Cheung.
Intelligent Database Systems Lab N.Y.U.S.T. I. M. Unsupervised word sense disambiguation for Korean through the acyclic weighted digraph using corpus and.
Intelligent Database Systems Lab N.Y.U.S.T. I. M. 1 Visualization of multi-algorithm clustering for better economic decisions - The case of car pricing.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Information Loss of the Mahalanobis Distance in High Dimensions-
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Multiclass boosting with repartitioning Graduate : Chen,
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 An initialization method to simultaneously find initial.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology O( ㏒ 2 M) Self-Organizing Map Algorithm Without Learning.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 A personal route prediction system base on trajectory.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 A self-organizing map for adaptive processing of structured.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Adaptive FIR Neural Model for Centroid Learning in Self-Organizing.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Cost- sensitive boosting for classification of imbalanced.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Direct mining of discriminative patterns for classifying.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Comparing Association Rules and Decision Trees for Disease.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Growing Hierarchical Tree SOM: An unsupervised neural.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Mining Advisor-Advisee Relationships from Research Publication.
Intelligent Database Systems Lab Advisor : Dr. Hsu Graduate : Yu Cheng Chen Author : Yongqiang Cao Jianhong Wu 國立雲林科技大學 National Yunlin University of Science.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Dual clustering : integrating data clustering over optimization.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Presenter : Chien-Shing Chen Author: Gustavo.
國立雲林科技大學 National Yunlin University of Science and Technology Mining Generalized Associations of Semantic Relations from Textual Web Content Tao Jiang,
Intelligent Database Systems Lab N.Y.U.S.T. I. M. An integrated scheme for feature selection and parameter setting in the support vector machine modeling.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 Prediction model building and feature selection with support.
Intelligent Database Systems Lab N.Y.U.S.T. I. M. Visualizing social network concepts Presenter : Chun-Ping Wu Authors :Bin Zhu, Stephanie Watts, Hsinchun.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Graduate : Chun Kai Chen Author : Andrew.
Intelligent Database Systems Lab N.Y.U.S.T. I. M. Named Entity Disambiguation by Leveraging Wikipedia Semantic Knowledge Presenter : Jiang-Shan Wang Authors.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology A Nonlinear Mapping for Data Structure Analysis John W.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology 1 f-information measures in medical image registration Presenter.
Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Advisor : Dr. Hsu Graduate : Yu Cheng Chen Author: Michael.
Presentation transcript:

Intelligent Database Systems Lab 國立雲林科技大學 National Yunlin University of Science and Technology Efficient Optimal Linear Boosting of a Pair of Classifiers Victor Boyarshinov and Malik Magdon-Ismail, IEEE Transaction on neural networks, Vol. 18, No. 2, 2007, pp Presenter : Wei-Shen Tai Advisor : Professor Chung-Chian Hsu 2007/5/16

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Outline Introduction Optimal fat separators in two dimension  Mirrored-radial coordinates  Maximizing the Margin Subject to Minimum Weight Leave-one-out error Discussions Comments

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Motivation MODEL aggregation  Enhancing the statistical performance of a set of weak classifiers to obtain a stronger classifier.  For example, boosting and bagging. Combinatorial problem of boosting a pair of classifiers  What is the optimal linear combination of this pair of classifiers?

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Objective Optimal linear separation (l)  the optimal linear boosted classifier for g 1 and g 2 classifier (for whatever optimal means) in terms of its corresponding optimal linear classifier in the 2-D space.

N.Y.U.S.T. I. M. Intelligent Database Systems Lab A (linearly) boosted classification function  A (linearly) boosted classification function g(z)  g(z) = w 0 +w 1 g 1 (z) + w 2 g 2 (z) The corresponding classifier is  sign(g(z)) = sign(w 0 +w 1 g 1 (z) + w 2 g 2 (z))  Each data point z i  d can be mapped onto this 2-D feature space by { z i, y i }  A linearly boosted classifier corresponds exactly to a linear separator in this 2-D space.

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Optimal linear separation in  2 Efficient algorithms for exact linear separation  Applicable to the case where the data points are not linearly separable. Definition 1.2: Two sets A and B are linearly separable iff  v, v 0 such that v T x + v 0 > 0,  x  A and v T x + v 0 < 0,  x  B. The pair (v, v 0 ) defines an (oriented) separating hyperplane. If two sets and are linearly separable, the margin of a separating hyperplane is the minimum distance of a data point to the hyperplane. The maximum margin separating hyperplane is a separating hyperplane with maximum possible margin. The weight (or error) for is the total weight summed over the points misclassified by  A separator l* is optimal if it has minimum weight

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Optimal fat separator Definition 1.3 (Optimal fat separator)  A hyperplane l = (v, v 0 ) is an optimal fat separator for A and B if it is optimal and is also a maximum margin separator for A’(l) and B’(l).  Analogously define the optimal fat separator with respect to the new separable set that would be obtained if instead of removing the misclassified points, we flip the classes of these points—all our results apply here as well. Definition 2.1: A separator set Q  A  B  A set with the following property: if the points in Q are deleted, the remaining points are linearly separable. Definition 2.4: For hyperplane l, the positive separator set Q + (l)  Contains all misclassified points except the positive points (in A) that lie on l.

N.Y.U.S.T. I. M. Intelligent Database Systems Lab The best positive separator hyperplane Lemma 2.6: The optimal positive separator set over all hyperplanes passing through a candidate central point. Mirrored-radial coordinate  A point s(x) as the projection of x onto the upper hemisphere of the unit circle, through the origin a +.  The mirrored-radial coordinate θ(x) is then the angle of s(x), i.e., θ (s(x)).  Find l with minimum weight

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Maximizing the margin subject to minimum weight Definition 2.12: For the set A  B, the margin of separator set Q,mar(Q), is the margin of the optimal fat separator for A  B \ Q.  Theorem 2.13: An optimal separator set Q* (1) For any other separator set W(Q*) ≦ W(Q), (2) and if W(Q*) =W(Q), then mar(Q*) ≧ mar(Q).  Convex hull Convex hull

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Convex hull Definition  A set of points X in a real vector space V is the minimal convex set containing X.  To show this exists, it is necessary to see that every X is contained in at least one convex set (the whole space V, for example), and any intersection of convex sets containing X is also a convex set containing X.

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Leave-one-out error Estimation of the accuracy of learning algorithms  Let e i denote the error of C (i) applied to the input point x i  Three types of x i Type I: x i is classified correctly by all distinct optimal fat separators constructed for. Such an makes no contribution to the leave-one-out error. Type II: x i is misclassified by all optimal fat separators constructed for X (i). Such an x i contributes w i to the leave-one-out error. Type III: There are distinct optimal separator sets for X (i). Let N c of these occurrences result in fat separators that classify correctly and N e of them misclassify.

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Discussions Optimal linear boosting of a pair of classification functions  A significant improvement over the brute force exponential algorithm.  A smarter way to enumerate these lines to result in a speed up of a factor close to n.  Extended to maximize the margin among all optimal separator set. Limitation and future work  Only applies to the 2-D case. This is a significant limitation for linear separation.  Extend the approach further to obtain an optimal boosting of an arbitrary number of classification functions.

N.Y.U.S.T. I. M. Intelligent Database Systems Lab Comments Advantage  A model aggregation method for combining different classifiers.  Mirrored-radial coordinates save the search space for finding optimal separator hyperplane. Drawback  It lacks extra diagrams to explain lemmas and theorems in detail.  There is no experiment to demonstrate the performance of this method with other related methods in this paper. Application  Model aggregation for classification related applications.