Transcription of Text by Incremental Support Vector machine Anurag Sahajpal and Terje Kristensen.

Slides:



Advertisements
Similar presentations
Introduction to Support Vector Machines (SVM)
Advertisements

ECG Signal processing (2)
Image classification Given the bag-of-features representations of images from different classes, how do we learn a model for distinguishing them?
SVM - Support Vector Machines A new classification method for both linear and nonlinear data It uses a nonlinear mapping to transform the original training.
ONLINE ARABIC HANDWRITING RECOGNITION By George Kour Supervised by Dr. Raid Saabne.
An Introduction of Support Vector Machine

An Introduction of Support Vector Machine
Support Vector Machines
SVM—Support Vector Machines
Support vector machine
Search Engines Information Retrieval in Practice All slides ©Addison Wesley, 2008.
Machine learning continued Image source:
CSCI 347 / CS 4206: Data Mining Module 07: Implementations Topic 03: Linear Models.
Discriminative and generative methods for bags of features
Support Vector Machine
Image classification Given the bag-of-features representations of images from different classes, how do we learn a model for distinguishing them?
Fuzzy Support Vector Machines (FSVMs) Weijia Wang, Huanren Zhang, Vijendra Purohit, Aditi Gupta.
Support Vector Machines MEDINFO 2004, T02: Machine Learning Methods for Decision Support and Discovery Constantin F. Aliferis & Ioannis Tsamardinos Discovery.
Reduced Support Vector Machine
Support Vector Machines Kernel Machines
CS 4700: Foundations of Artificial Intelligence
What is Learning All about ?  Get knowledge of by study, experience, or being taught  Become aware by information or from observation  Commit to memory.
Face Processing System Presented by: Harvest Jang Group meeting Fall 2002.
Statistical Learning Theory: Classification Using Support Vector Machines John DiMona Some slides based on Prof Andrew Moore at CMU:
Optimization Theory Primal Optimization Problem subject to: Primal Optimal Value:
Review Rong Jin. Comparison of Different Classification Models  The goal of all classifiers Predicating class label y for an input x Estimate p(y|x)
An Introduction to Support Vector Machines Martin Law.
Classification III Tamara Berg CS Artificial Intelligence Many slides throughout the course adapted from Svetlana Lazebnik, Dan Klein, Stuart Russell,
Efficient Model Selection for Support Vector Machines
Support Vector Machine & Image Classification Applications
CSE 473/573 Computer Vision and Image Processing (CVIP) Ifeoma Nwogu Lecture 24 – Classifiers 1.
Support Vector Machines Mei-Chen Yeh 04/20/2010. The Classification Problem Label instances, usually represented by feature vectors, into one of the predefined.
1 SUPPORT VECTOR MACHINES İsmail GÜNEŞ. 2 What is SVM? A new generation learning system. A new generation learning system. Based on recent advances in.
计算机学院 计算感知 Support Vector Machines. 2 University of Texas at Austin Machine Learning Group 计算感知 计算机学院 Perceptron Revisited: Linear Separators Binary classification.
1 CSC 4510, Spring © Paula Matuszek CSC 4510 Support Vector Machines 2 (SVMs)
Seungchan Lee Intelligent Electronic Systems Human and Systems Engineering Department of Electrical and Computer Engineering Software Release and Support.
Machine Learning Using Support Vector Machines (Paper Review) Presented to: Prof. Dr. Mohamed Batouche Prepared By: Asma B. Al-Saleh Amani A. Al-Ajlan.
Kernel Methods A B M Shawkat Ali 1 2 Data Mining ¤ DM or KDD (Knowledge Discovery in Databases) Extracting previously unknown, valid, and actionable.
SVM Support Vector Machines Presented by: Anas Assiri Supervisor Prof. Dr. Mohamed Batouche.
Support Vector Machines Reading: Ben-Hur and Weston, “A User’s Guide to Support Vector Machines” (linked from class web page)
Jun-Won Suh Intelligent Electronic Systems Human and Systems Engineering Department of Electrical and Computer Engineering Speaker Verification System.
Classifiers Given a feature representation for images, how do we learn a model for distinguishing features from different classes? Zebra Non-zebra Decision.
An Introduction to Support Vector Machines (M. Law)
Using Support Vector Machines to Enhance the Performance of Bayesian Face Recognition IEEE Transaction on Information Forensics and Security Zhifeng Li,
Learning Vilanova Multi-classification by using tri- class SVM Luis González and Francisco Velasco Appl. Economics – University of Sevilla Cecilio.
1 Chapter 6. Classification and Prediction Overview Classification algorithms and methods Decision tree induction Bayesian classification Lazy learning.
Artificial Intelligence Research Laboratory Bioinformatics and Computational Biology Program Computational Intelligence, Learning, and Discovery Program.
CS 1699: Intro to Computer Vision Support Vector Machines Prof. Adriana Kovashka University of Pittsburgh October 29, 2015.
University of Texas at Austin Machine Learning Group Department of Computer Sciences University of Texas at Austin Support Vector Machines.
Dec 21, 2006For ICDM Panel on 10 Best Algorithms Support Vector Machines: A Survey Qiang Yang, for ICDM 2006 Panel Partially.
Support Vector Machines. Notation Assume a binary classification problem. –Instances are represented by vector x   n. –Training examples: x = (x 1,
Final Exam Review CS479/679 Pattern Recognition Dr. George Bebis 1.
Goal of Learning Algorithms  The early learning algorithms were designed to find such an accurate fit to the data.  A classifier is said to be consistent.
Support Vector Machines (SVM): A Tool for Machine Learning Yixin Chen Ph.D Candidate, CSE 1/10/2002.
Chapter 6. Classification and Prediction Classification by decision tree induction Bayesian classification Rule-based classification Classification by.
Chapter 6. Classification and Prediction Classification by decision tree induction Bayesian classification Rule-based classification Classification by.
Feature Selction for SVMs J. Weston et al., NIPS 2000 오장민 (2000/01/04) Second reference : Mark A. Holl, Correlation-based Feature Selection for Machine.
CS 2750: Machine Learning Support Vector Machines Prof. Adriana Kovashka University of Pittsburgh February 17, 2016.
Chapter 10 The Support Vector Method For Estimating Indicator Functions Intelligent Information Processing Laboratory, Fudan University.
Incremental Reduced Support Vector Machines Yuh-Jye Lee, Hung-Yi Lo and Su-Yun Huang National Taiwan University of Science and Technology and Institute.
1 Kernel Machines A relatively new learning methodology (1992) derived from statistical learning theory. Became famous when it gave accuracy comparable.
Roughly overview of Support vector machines Reference: 1.Support vector machines and machine learning on documents. Christopher D. Manning, Prabhakar Raghavan.
A Brief Introduction to Support Vector Machine (SVM) Most slides were from Prof. A. W. Moore, School of Computer Science, Carnegie Mellon University.
Neural networks and support vector machines
CS 9633 Machine Learning Support Vector Machines
MIRA, SVM, k-NN Lirong Xia. MIRA, SVM, k-NN Lirong Xia.
CS 2750: Machine Learning Support Vector Machines
The following slides are taken from:
MIRA, SVM, k-NN Lirong Xia. MIRA, SVM, k-NN Lirong Xia.
Presentation transcript:

Transcription of Text by Incremental Support Vector machine Anurag Sahajpal and Terje Kristensen

Outline Introduction Theory of Incremental SVM Application Discussion, further work and references

Introduction Phoneme : the basic abstract symbol representing speech sound Transcription : process of converting text (word) into corresponding phonetic sequence Letter-to-phoneme correspondence is generally not one-to-one Examples : ”lønnsoppgaver” trascribes to ”natt” transcribes to nAt while rar to rA:r

The Problem Phoneme transcription an instance of more general problem of Pattern recognition Phonetic rules compiled by experts are time consuming and fixed for a particular langauge What is required is an automatic approach, independent of any particular language

The Problem Machine learning approach using SVM reported in earlier paper The phonemic data in a language shows regional variation Distributed learning by SVM may be tried to adapt to geografically distributed phonemic data

Support Vector Machine Distribution free Non-parametric Non-linear High-dimensional Small training data sets Convex QP problem Good generalization performance Support Vectors Margin Width x2x2 x1x1

Support Vector Machine In a nutshell: map the data to a predetermined very high- dimensional space via a kernel function Find the hyperplane that maximizes the margin between the two classes If data are not separable find the hyperplane that maximizes the margin and minimizes the (a weighted average of the) misclassifications

Which Separating Hyperplane to Use? Var 1 Var 2

Maximizing the Margin Var 1 Var 2 Margin Width IDEA 1: Select the separating hyperplane that maximizes the margin!

MultiClass SVMs One-versus-all Train n binary classifiers, one for each class against all other classes. Predicted class is the class of the most confident classifier One-versus-one Train n(n-1)/2 classifiers, each discriminating between a pair of classes Several strategies for selecting the final classification based on the output of the binary SVMs

Outline Theory of Incremental SVM

SVM in Incremental and Distributed Settings Performance constriants with SVM training for large-scale problems Cumulative learning algorithms that incorporate new data over time (incremental) and space (distributed) Modifications to batch SVM learning to adapt to cumulative settings Calls for provable convergence properties

A naive approach to cumulative learning SVM learns D 1 and generate a set of support vectors SV 1 add SV 1 to D 2 to get a data set D ` 2 SVM learns D ` 2 and generate a set of support vectors SV 2

Incremental SVM Learning Convex hull of a set of points, S, is the smallest convex set containing S U-Closure property satisfied for convex hulls Vconv(Vconv(A 1 ) U A 2 ) = Vconv(A 1 U A 2 ) where Vconv(A) denote the vertices of a convex hull of a set A

Incremental SVM Learning learning algorithm, L, computes Vconv(D 1 (+) ) and Vconv(D 1 (-) ) Add Vconv(D 1 (+) ) to D 2 (+) to obtain D` 2 (+) Add Vconv(D 1 (-) ) to D 2 (-) to obtain D` 2 (-) L computes Vconv(D` 2 (+) ) and Vconv(D` 2 (-) ) Generate a training: D 12 = Vconv(D` 2 (+) ) U Vconv(D` 2 (-) ) compute SVM (D 12 )

Outline Application

SAMPA for Norwegian SAMPA (Speech Assessment Methods Phonetic Alphabet) - A computer readable phonetic alphabet Consonants and Vowels are classified into different subgroups : Consonants – plosives(6), fricatives(7), sonorant consonants(5) Vowels – long(9), short(9), Diphthongs(7) In our work, an estimate of 43 phonemes plus 10 additional phonetic symbols

Example of Training data file Some examples of transcription of words using the Sampa notation: WordsTranscription

Transcription Method Each letter pattern is a window onto a segment of the word where the phoneme to be predicted is in the middle of the window The size of the window is selected to 7 letters in all the experiments * e l e v e n context active

Pre-processing and coding A pattern file of data consist of words and their trancription Each pattern file is preprocessed before it is fed into SVM An internal coding table is defined in the program to represent each letter and its corresponding phoneme Example data file for LIBSVM

0 4:52 5:51 6:38 7:51 0 3:52 4:51 5:38 6:51 7:37 0 2:52 3:51 4:38 5:51 6:37 0 1:52 2:51 3:38 4:51 5:37 0 1:51 2:38 3:51 4:37 1 4:55 5:54 6:53 7:55 0 3:55 4:54 5:53 6:55 0 2:55 3:54 4:53 5:55 0 1:55 2:54 3:53 4:55 0 4:55 5:54 6:53 7:51

Experiment Various steps in the experiment One-versus-all training patterns Generation of 54 class files Separate training for 54 corresponding models

Experiment Various steps in the experiment The test file containing patterns is tested by each model and voting was carried out The output file and the true output was compared to find the accuracy

Outline Discussion, further work and references

Discussion and Future Work Complexity of convex hull computations have an exponential dependence on the dimensionality of the feature space. Implementation and modification to the standard batch-mode SVM to incorporate convex hull algorithm Extension to non-linear SVM classifier

References Caragea D. and Silvescu A and Honavar V “Agents that learn from distributed data sources” In fourth International Conference on Autonomous Agents C. J. C. Burges. A Tutorial on Support Vector Machines for Pattern Recognition. Knowledge Discovery and Data Mining, 2(2), 1998.