Music instrument categorization using multilayer perceptron network Ivana Andjelkovic PHY 171, Winter 2011

Size: px
Start display at page:

Download "Music instrument categorization using multilayer perceptron network Ivana Andjelkovic PHY 171, Winter 2011"

Transcription

1 Music instrument categorization using multilayer perceptron network Ivana Andjelkovic PHY 171, Winter 2011 Abstract Audio content description is one of the key components to multimedia search, classification and source identification. This project examines precision of audio content description based on spectral information only on one hand, and spectral and temporal on the other. Multilayer neural network with varied parameters is used to classify musical instruments based on obtained descriptors. 1. Introduction 1.1. Motivation Research of music instrument recognition methods has received much attention and its numerous applications range from automatic music cataloguing, transcription, audio indexing to identification of performers based on their instrument playing style. The goal of this project is to gain practical knowledge of a particular classification method, namely Multilayer Perceptron and apply it towards classifying musical instruments based on two different sound analysis approaches. Music instrument recognition most often consists of two phases: (1) Audio content analysis musical features can be temporal, spectral and perceptual, to name a few, and some of the commonly used computational feature extraction methods include Mel-frequency cepstral coefficients (MFCC), Short Time Fourier Transform (STFT) and Wavelet transforms. (2) Classification various learning methods that analyze data and recognize patterns, such as neural networks training and support vector machines algorithms, 1

2 have been used independently with different success rates. The combination of one or more different methods may be applied towards solving a classification problem, which can yield better results [1]. This project utilizes FFT and Wavelet transform techniques to analyze audio content, and multilayer neural network to classify it Sound analysis It has been shown by Brown et al. [5], Dubnov et al. [6], and others that spectral information is sufficient to distinguish between instruments from different families. Most commonly used algorithm to obtain the spectrum of audio signal is Fast Fourier Transform (FFT). However, sounds that musical instruments produce vary in how their features, such as loudness and spectral content change over time. For instance, "attack" and "decay" (Figure 1) of a sound have a great effect on the instrument's sonic character (Figure 2). Figure 1: Sound envelope; Attack time is the time taken for initial run-up of level from nil to peak, beginning when the key is first pressed. Decay time is the time taken for the subsequent run down from the attack level to the designated sustain level. Sustain level is the level during the main sequence of the sound's duration, until the key is released. Release time is the time taken for the level to decay from the sustain level to zero after the key is released. 2

3 1500 Strings x Piano 1 x x 10 4 x 10 4 Figure 2: Spectrum obtained using FFT and plotted in Matlab (left), and sound wave (right) of two second isolated note played on orchestral strings (top) and piano (bottom). Notice the difference between almost instantaneous attack of piano sound and gradual attack for strings. There exist numerous techniques for observing how spectrum of a signal changes over time. Perhaps the best known among them is Short-time Fourier Transform (STFT). The data to be transform is broken up into windows of desired size, and FFT is applied to each one of them. The drawback of STFT is that it has fixed resolution all windows are the same size. The width of the windowing function relates to how the signal is represented. It determines whether there is good frequency resolution (frequency components close together can be separated) or good time resolution (the time at which frequencies change). A wide window gives better frequency resolution but poor time resolution. A narrower window gives good time resolution but poor frequency resolution. The Wavelet Transform overcomes the fixed size windowing problem and gives good time resolution for high-frequency events, and good frequency resolution for lowfrequency events. Thus it yields better results in examining most real signals. 3

4 2. Experiment 2.1. Problem description The experiment examines the precision of classifying music instruments from different families and within a same family based on spectral information only and based on spectral and temporal information. (Figure 3) Figure 3 As stated in the introduction, numerous studies showed that spectral information can be sufficient to categorize instruments into different families. In particular, study by Hassan [4] confirms this finding by examining isolated notes played on different musical instruments. This study served as a guide for designing the experiment. Several studies addressed usage of wavelet transform and neural networks to classify musical instruments [1],[2],[3]. For instance, it has been shown that wavelet coefficients used to train a neural network yield 78% success rate in identifying unknown complex classical movements [2] and can be successfully used to identify dominant instrument in a short clip of music [3]. Based on the successful classification results of complex audio data using wavelet transform and neural networks, I hypothesized the same methods would work well in classification of isolated notes played on instruments in the same family Training data processing Due to time constrains, the data was obtained by recording two second isolated notes, played on software instruments in Mac OS application Garage Band. Although the sounds are realistic, they are synthesized rather than obtained by recording real instruments. In addition, the sustain and release rates of different instruments may not be proportional across different instruments. Four instruments belonging to different families are flute, piano, guitar and orchestral strings. Three instruments belonging to wind family are flute, saxophone and trumpet. Selection of the instruments was solely based on their availability. Four octaves, totaling 48 notes were obtained for each instrument, at Hz sampling rate. 4

5 Results of both FFT and Wavelet analysis were relatively large and too cumbersome for training a neural network. Therefore, an important step is to reduce the size of data while maintaining the unique representation of each sample. In the case of FFT analysis, 1024-point FFT was applied to each note. Next, second half of result was discarded because of the symmetry. The result was a vector of 512 elements for each note. In the case of Wavelet transform, a technique Wavelet Rank Dispersion Vectors (WRDV) was used to reduce the data size. WRDV is defined as a histogram of the rank orders obtained by the wavelet coefficients of a given wavelet scale among all the coefficients. First, continuous wavelet transform using Meyer wavelets was applied to the original data. Second, a large number of resulting coefficients were represented as rank vectors containing 530 elements Classification method Neural Networks Matlab toolbox was used to train the multilayer network using various parameters and easily plot results of experiment. Among several available methods and learning algorithms, the one used to address classification problem was Neural network pattern recognition tool. In the case of classification of instruments from different families, there were either 530 (WRDV) or 512 (FFT) input neurons, and 4 output neurons representing 4 instrument. There were 3 instruments belonging to the same family, and so the number of output neurons in such case was 3. Number of neurons in the hidden layer was varied throughout the experiment. Following are the default settings for some of the training functions that were not changed during the experiment: (1) Transfer function: tan-sigmoid for hidden layers, and linear for output layer. (2) Backpropagation network training function: Scaled conjugate gradient backpropagation (3) Backpropagation weight/bias learning function: Gradient descent with momentum weight and bias learning function The pattern recognition tool by default randomly divides the input data set such that 70% of data is used for training, 15% for validation and remaining 15% of data for testing. 5

6 2.4 Results Different families of instruments The analysis of results was not conducted in great detail. However, it was sufficient to make general conclusion about efficiency of different approaches. As expected, classification of instruments from different families based on spectral information only was successful, but inferior to classification based on wavelet coefficients. Figure 4 shows Receiver Operating Characteristic 1 plot, and confusion matrices 2 for three different training sessions based on FFT data. The number of hidden neurons is 40, which yielded the best classification results. The average Mean Square Error for all three sessions was Further analysis may answer the question which instrument was least or most often misclassified. Figure 4: ROC plots and Confusion matrices for three different sessions on training with FFT data. Number of hidden neurons is True positive rate vs. false positive rate for a binary classifier system 2 Each column of the matrix represents the instances in a predicted class, while each row represents the instances in an actual class 6

7 Figure 5 shows the ROC and confusion matrix of the most successful training session using WRDV, where hidden layer contains 30 neurons. Average Mean Squared Error or two different sessions was Figure 5: The most successful session based on WRDV data Wind family of instruments Distinguishing among instruments within the same family was not satisfactory based on FFT data but, as hypothesized, it was successful based on WRDV data. Figure 6 shows ROC plots and Confusion matrices for three different training sessions based on FFT data only, where number of neurons in the hidden layer was 30. Average MSE for all sessions was Neither increasing the number of hidden neurons, nor changing the division of data set such that 80% is used for training, significantly improved the results. 7

8 Figure 6: Results of three training sessions based on FFT data Figure 7 shows results of the most successful training session based on WRDV data, where 50 hidden neurons were employed. Mean Squared Error was Although the results of other training sessions are not available at the moment, they were overall satisfactory. Figure 7: Results of the most successful training session based on WRDV, utilizing 50 hidden neurons 8

9 3. Conclusion Results of the experiment show that Wavelet Rank Dispersion Vector measure used to train multilayer neural network can be successfully employed to classify solo instruments. The experiment was designed and conducted within limited amount of time, and it could be greatly improved for further studies. Nevertheless, learning experience was invaluable and obtained results align with the findings in the field. First, the loudness and release time of sound envelopes should be normalized across samples for all instruments. Second, different degrees of similarity should be noted between instruments within the same family. This experiment examined three instruments within wind family, but the classification method can be, for instance, tested for woodwind or brass instruments. Third, the choice of different backpropagation algorithm may influence the success rate of classification. Finally, method can be compared against others in classification of more complex musical excerpts. References: 9

10 [1] Identifying the classical music composition of an unknown performance with wavelet dispersion vector and neural nets, Stephan Rein; Martin Reisslein [2] Audio content description with wavelets and neural nets, Stephan Rein, Martin Reisslein, and Thomas Sikora [3] Musical Instrument Identification Using Wavelets and Neural Networks, Jeffrey Livingston, Nathan Shepard [4] Instruments recognition using neural networks and spectral information, Ezzaidi Hassan [5] Feature dependence in the automatic identification of musical woodwind instruments, Brown J. C., Houix O. and McAdams S., [6] Polyspectra as measures of sound texture and timbre, Dubnov S., Tishby N. and Cohen D 10

MUSICAL INSTRUMENT FAMILY CLASSIFICATION

MUSICAL INSTRUMENT FAMILY CLASSIFICATION MUSICAL INSTRUMENT FAMILY CLASSIFICATION Ricardo A. Garcia Media Lab, Massachusetts Institute of Technology 0 Ames Street Room E5-40, Cambridge, MA 039 USA PH: 67-53-0 FAX: 67-58-664 e-mail: rago @ media.

More information

Auto-Tuning Using Fourier Coefficients

Auto-Tuning Using Fourier Coefficients Auto-Tuning Using Fourier Coefficients Math 56 Tom Whalen May 20, 2013 The Fourier transform is an integral part of signal processing of any kind. To be able to analyze an input signal as a superposition

More information

Artificial Neural Network for Speech Recognition

Artificial Neural Network for Speech Recognition Artificial Neural Network for Speech Recognition Austin Marshall March 3, 2005 2nd Annual Student Research Showcase Overview Presenting an Artificial Neural Network to recognize and classify speech Spoken

More information

This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore.

This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore. This document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore. Title Transcription of polyphonic signals using fast filter bank( Accepted version ) Author(s) Foo, Say Wei;

More information

Quarterly Progress and Status Report. Measuring inharmonicity through pitch extraction

Quarterly Progress and Status Report. Measuring inharmonicity through pitch extraction Dept. for Speech, Music and Hearing Quarterly Progress and Status Report Measuring inharmonicity through pitch extraction Galembo, A. and Askenfelt, A. journal: STL-QPSR volume: 35 number: 1 year: 1994

More information

Short-time FFT, Multi-taper analysis & Filtering in SPM12

Short-time FFT, Multi-taper analysis & Filtering in SPM12 Short-time FFT, Multi-taper analysis & Filtering in SPM12 Computational Psychiatry Seminar, FS 2015 Daniel Renz, Translational Neuromodeling Unit, ETHZ & UZH 20.03.2015 Overview Refresher Short-time Fourier

More information

Speech Signal Processing: An Overview

Speech Signal Processing: An Overview Speech Signal Processing: An Overview S. R. M. Prasanna Department of Electronics and Electrical Engineering Indian Institute of Technology Guwahati December, 2012 Prasanna (EMST Lab, EEE, IITG) Speech

More information

A simple application of Artificial Neural Network to cloud classification

A simple application of Artificial Neural Network to cloud classification A simple application of Artificial Neural Network to cloud classification Tianle Yuan For AOSC 630 (by Prof. Kalnay) Introduction to Pattern Recognition (PR) Example1: visual separation between the character

More information

Teaching Fourier Analysis and Wave Physics with the Bass Guitar

Teaching Fourier Analysis and Wave Physics with the Bass Guitar Teaching Fourier Analysis and Wave Physics with the Bass Guitar Michael Courtney Department of Chemistry and Physics, Western Carolina University Norm Althausen Lorain County Community College This article

More information

Analecta Vol. 8, No. 2 ISSN 2064-7964

Analecta 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 information

Separation and Classification of Harmonic Sounds for Singing Voice Detection

Separation and Classification of Harmonic Sounds for Singing Voice Detection Separation and Classification of Harmonic Sounds for Singing Voice Detection Martín Rocamora and Alvaro Pardo Institute of Electrical Engineering - School of Engineering Universidad de la República, Uruguay

More information

A Sound Analysis and Synthesis System for Generating an Instrumental Piri Song

A Sound Analysis and Synthesis System for Generating an Instrumental Piri Song , pp.347-354 http://dx.doi.org/10.14257/ijmue.2014.9.8.32 A Sound Analysis and Synthesis System for Generating an Instrumental Piri Song Myeongsu Kang and Jong-Myon Kim School of Electrical Engineering,

More information

Artificial 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 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 information

Chapter 4: Artificial Neural Networks

Chapter 4: Artificial Neural Networks Chapter 4: Artificial Neural Networks CS 536: Machine Learning Littman (Wu, TA) Administration icml-03: instructional Conference on Machine Learning http://www.cs.rutgers.edu/~mlittman/courses/ml03/icml03/

More information

Programming Exercise 3: Multi-class Classification and Neural Networks

Programming 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 information

Feed-Forward mapping networks KAIST 바이오및뇌공학과 정재승

Feed-Forward mapping networks KAIST 바이오및뇌공학과 정재승 Feed-Forward mapping networks KAIST 바이오및뇌공학과 정재승 How much energy do we need for brain functions? Information processing: Trade-off between energy consumption and wiring cost Trade-off between energy consumption

More information

An Introduction to Neural Networks

An Introduction to Neural Networks An Introduction to Vincent Cheung Kevin Cannons Signal & Data Compression Laboratory Electrical & Computer Engineering University of Manitoba Winnipeg, Manitoba, Canada Advisor: Dr. W. Kinsner May 27,

More information

L9: Cepstral analysis

L9: Cepstral analysis L9: Cepstral analysis The cepstrum Homomorphic filtering The cepstrum and voicing/pitch detection Linear prediction cepstral coefficients Mel frequency cepstral coefficients This lecture is based on [Taylor,

More information

Annotated bibliographies for presentations in MUMT 611, Winter 2006

Annotated bibliographies for presentations in MUMT 611, Winter 2006 Stephen Sinclair Music Technology Area, McGill University. Montreal, Canada Annotated bibliographies for presentations in MUMT 611, Winter 2006 Presentation 4: Musical Genre Similarity Aucouturier, J.-J.

More information

Little LFO. Little LFO. User Manual. by Little IO Co.

Little LFO. Little LFO. User Manual. by Little IO Co. 1 Little LFO User Manual Little LFO by Little IO Co. 2 Contents Overview Oscillator Status Switch Status Light Oscillator Label Volume and Envelope Volume Envelope Attack (ATT) Decay (DEC) Sustain (SUS)

More information

Electroencephalography Analysis Using Neural Network and Support Vector Machine during Sleep

Electroencephalography Analysis Using Neural Network and Support Vector Machine during Sleep Engineering, 23, 5, 88-92 doi:.4236/eng.23.55b8 Published Online May 23 (http://www.scirp.org/journal/eng) Electroencephalography Analysis Using Neural Network and Support Vector Machine during Sleep JeeEun

More information

1. Classification problems

1. 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 information

FOURIER TRANSFORM BASED SIMPLE CHORD ANALYSIS. UIUC Physics 193 POM

FOURIER TRANSFORM BASED SIMPLE CHORD ANALYSIS. UIUC Physics 193 POM FOURIER TRANSFORM BASED SIMPLE CHORD ANALYSIS Fanbo Xiang UIUC Physics 193 POM Professor Steven M. Errede Fall 2014 1 Introduction Chords, an essential part of music, have long been analyzed. Different

More information

Using Artifical Neural Networks to Model Opponents in Texas Hold'em

Using Artifical Neural Networks to Model Opponents in Texas Hold'em CMPUT 499 - Research Project Review Using Artifical Neural Networks to Model Opponents in Texas Hold'em Aaron Davidson email: davidson@cs.ualberta.ca November 28th, 1999 Abstract: This paper describes

More information

Automatic Detection of Emergency Vehicles for Hearing Impaired Drivers

Automatic Detection of Emergency Vehicles for Hearing Impaired Drivers Automatic Detection of Emergency Vehicles for Hearing Impaired Drivers Sung-won ark and Jose Trevino Texas A&M University-Kingsville, EE/CS Department, MSC 92, Kingsville, TX 78363 TEL (36) 593-2638, FAX

More information

Voltage. Oscillator. Voltage. Oscillator

Voltage. Oscillator. Voltage. Oscillator fpa 147 Week 6 Synthesis Basics In the early 1960s, inventors & entrepreneurs (Robert Moog, Don Buchla, Harold Bode, etc.) began assembling various modules into a single chassis, coupled with a user interface

More information

Efficient online learning of a non-negative sparse autoencoder

Efficient online learning of a non-negative sparse autoencoder and Machine Learning. Bruges (Belgium), 28-30 April 2010, d-side publi., ISBN 2-93030-10-2. Efficient online learning of a non-negative sparse autoencoder Andre Lemme, R. Felix Reinhart and Jochen J. Steil

More information

Automatic Evaluation Software for Contact Centre Agents voice Handling Performance

Automatic Evaluation Software for Contact Centre Agents voice Handling Performance International Journal of Scientific and Research Publications, Volume 5, Issue 1, January 2015 1 Automatic Evaluation Software for Contact Centre Agents voice Handling Performance K.K.A. Nipuni N. Perera,

More information

Sound Quality Evaluation of Hermetic Compressors Using Artificial Neural Networks

Sound Quality Evaluation of Hermetic Compressors Using Artificial Neural Networks Purdue University Purdue e-pubs International Compressor Engineering Conference School of Mechanical Engineering 2006 Sound Quality Evaluation of Hermetic Compressors Using Artificial Neural Networks Claudio

More information

Establishing the Uniqueness of the Human Voice for Security Applications

Establishing the Uniqueness of the Human Voice for Security Applications Proceedings of Student/Faculty Research Day, CSIS, Pace University, May 7th, 2004 Establishing the Uniqueness of the Human Voice for Security Applications Naresh P. Trilok, Sung-Hyuk Cha, and Charles C.

More information

Advanced analytics at your hands

Advanced 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 information

Blood Vessel Classification into Arteries and Veins in Retinal Images

Blood Vessel Classification into Arteries and Veins in Retinal Images Blood Vessel Classification into Arteries and Veins in Retinal Images Claudia Kondermann and Daniel Kondermann a and Michelle Yan b a Interdisciplinary Center for Scientific Computing (IWR), University

More information

Matlab GUI for WFB spectral analysis

Matlab GUI for WFB spectral analysis Matlab GUI for WFB spectral analysis Jan Nováček Department of Radio Engineering K13137, CTU FEE Prague Abstract In the case of the sound signals analysis we usually use logarithmic scale on the frequency

More information

Practical Design of Filter Banks for Automatic Music Transcription

Practical Design of Filter Banks for Automatic Music Transcription Practical Design of Filter Banks for Automatic Music Transcription Filipe C. da C. B. Diniz, Luiz W. P. Biscainho, and Sergio L. Netto Federal University of Rio de Janeiro PEE-COPPE & DEL-Poli, POBox 6854,

More information

ANN Based Fault Classifier and Fault Locator for Double Circuit Transmission Line

ANN 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 information

The Calculation of G rms

The Calculation of G rms The Calculation of G rms QualMark Corp. Neill Doertenbach The metric of G rms is typically used to specify and compare the energy in repetitive shock vibration systems. However, the method of arriving

More information

CHAPTER VII CONCLUSIONS

CHAPTER VII CONCLUSIONS CHAPTER VII CONCLUSIONS To do successful research, you don t need to know everything, you just need to know of one thing that isn t known. -Arthur Schawlow In this chapter, we provide the summery of the

More information

Appendices. Appendix A: List of Publications

Appendices. Appendix A: List of Publications Appendices Appendix A: List of Publications The following papers highlight the findings of this research. These articles were published in reputed journals during the course of this research program. 1.

More information

Digital image processing

Digital image processing 746A27 Remote Sensing and GIS Lecture 4 Digital image processing Chandan Roy Guest Lecturer Department of Computer and Information Science Linköping University Digital Image Processing Most of the common

More information

Graham s Guide to Synthesizers (part 1) Analogue Synthesis

Graham s Guide to Synthesizers (part 1) Analogue Synthesis Graham s Guide to Synthesizers (part ) Analogue Synthesis Synthesizers were originally developed to imitate or synthesise the sounds of acoustic instruments electronically. Early synthesizers used analogue

More information

HOG AND SUBBAND POWER DISTRIBUTION IMAGE FEATURES FOR ACOUSTIC SCENE CLASSIFICATION. Victor Bisot, Slim Essid, Gaël Richard

HOG AND SUBBAND POWER DISTRIBUTION IMAGE FEATURES FOR ACOUSTIC SCENE CLASSIFICATION. Victor Bisot, Slim Essid, Gaël Richard HOG AND SUBBAND POWER DISTRIBUTION IMAGE FEATURES FOR ACOUSTIC SCENE CLASSIFICATION Victor Bisot, Slim Essid, Gaël Richard Institut Mines-Télécom, Télécom ParisTech, CNRS LTCI, 37-39 rue Dareau, 75014

More information

Neural Networks and Support Vector Machines

Neural 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 information

FTIR Instrumentation

FTIR Instrumentation FTIR Instrumentation Adopted from the FTIR lab instruction by H.-N. Hsieh, New Jersey Institute of Technology: http://www-ec.njit.edu/~hsieh/ene669/ftir.html 1. IR Instrumentation Two types of instrumentation

More information

Adding Sinusoids of the Same Frequency. Additive Synthesis. Spectrum. Music 270a: Modulation

Adding Sinusoids of the Same Frequency. Additive Synthesis. Spectrum. Music 270a: Modulation Adding Sinusoids of the Same Frequency Music 7a: Modulation Tamara Smyth, trsmyth@ucsd.edu Department of Music, University of California, San Diego (UCSD) February 9, 5 Recall, that adding sinusoids of

More information

6.2.8 Neural networks for data mining

6.2.8 Neural networks for data mining 6.2.8 Neural networks for data mining Walter Kosters 1 In many application areas neural networks are known to be valuable tools. This also holds for data mining. In this chapter we discuss the use of neural

More information

Musical Analysis and Synthesis in Matlab

Musical Analysis and Synthesis in Matlab 3. James Stewart, Calculus (5th ed.), Brooks/Cole, 2003. 4. TI-83 Graphing Calculator Guidebook, Texas Instruments,1995. Musical Analysis and Synthesis in Matlab Mark R. Petersen (mark.petersen@colorado.edu),

More information

Novelty Detection in image recognition using IRF Neural Networks properties

Novelty Detection in image recognition using IRF Neural Networks properties Novelty Detection in image recognition using IRF Neural Networks properties Philippe Smagghe, Jean-Luc Buessler, Jean-Philippe Urban Université de Haute-Alsace MIPS 4, rue des Frères Lumière, 68093 Mulhouse,

More information

Doppler. Doppler. Doppler shift. Doppler Frequency. Doppler shift. Doppler shift. Chapter 19

Doppler. Doppler. Doppler shift. Doppler Frequency. Doppler shift. Doppler shift. Chapter 19 Doppler Doppler Chapter 19 A moving train with a trumpet player holding the same tone for a very long time travels from your left to your right. The tone changes relative the motion of you (receiver) and

More information

Recurrent Neural Networks

Recurrent Neural Networks Recurrent Neural Networks Neural Computation : Lecture 12 John A. Bullinaria, 2015 1. Recurrent Neural Network Architectures 2. State Space Models and Dynamical Systems 3. Backpropagation Through Time

More information

Neural Network Toolbox

Neural Network Toolbox Neural Network Toolbox A Tutorial for the Course Computational Intelligence http://www.igi.tugraz.at/lehre/ci Stefan Häusler Institute for Theoretical Computer Science Inffeldgasse 16b/I Abstract This

More information

B3. Short Time Fourier Transform (STFT)

B3. Short Time Fourier Transform (STFT) B3. Short Time Fourier Transform (STFT) Objectives: Understand the concept of a time varying frequency spectrum and the spectrogram Understand the effect of different windows on the spectrogram; Understand

More information

IBM SPSS Neural Networks 22

IBM SPSS Neural Networks 22 IBM SPSS Neural Networks 22 Note Before using this information and the product it supports, read the information in Notices on page 21. Product Information This edition applies to version 22, release 0,

More information

Due Wednesday, December 2. You can submit your answers to the analytical questions via either

Due Wednesday, December 2. You can submit your answers to the analytical questions via either ELEN 4810 Homework 6 Due Wednesday, December 2. You can submit your answers to the analytical questions via either - Hardcopy submission at the beginning of class on Wednesday, December 2, or - Electronic

More information

Emotion Detection from Speech

Emotion Detection from Speech Emotion Detection from Speech 1. Introduction Although emotion detection from speech is a relatively new field of research, it has many potential applications. In human-computer or human-human interaction

More information

Lecture 8 February 4

Lecture 8 February 4 ICS273A: Machine Learning Winter 2008 Lecture 8 February 4 Scribe: Carlos Agell (Student) Lecturer: Deva Ramanan 8.1 Neural Nets 8.1.1 Logistic Regression Recall the logistic function: g(x) = 1 1 + e θt

More information

THE SPECTRAL MODELING TOOLBOX: A SOUND ANALYSIS/SYNTHESIS SYSTEM. A Thesis. Submitted to the Faculty

THE SPECTRAL MODELING TOOLBOX: A SOUND ANALYSIS/SYNTHESIS SYSTEM. A Thesis. Submitted to the Faculty THE SPECTRAL MODELING TOOLBOX: A SOUND ANALYSIS/SYNTHESIS SYSTEM A Thesis Submitted to the Faculty in partial fulfillment of the requirements for the degree of Master of Arts in ELECTRO-ACOUSTIC MUSIC

More information

Modeling Affective Content of Music: A Knowledge Base Approach

Modeling Affective Content of Music: A Knowledge Base Approach Modeling Affective Content of Music: A Knowledge Base Approach António Pedro Oliveira, Amílcar Cardoso Centre for Informatics and Systems of the University of Coimbra, Coimbra, Portugal Abstract The work

More information

Estimation of Loudness by Zwicker's Method

Estimation of Loudness by Zwicker's Method Estimation of Loudness by Zwicker's Method Loudness is one category in the list of human perceptions of sound. There are many methods of estimating Loudness using objective measurements. No method is perfect.

More information

The Sonometer The Resonant String and Timbre Change after plucking

The Sonometer The Resonant String and Timbre Change after plucking The Sonometer The Resonant String and Timbre Change after plucking EQUIPMENT Pasco sonometers (pick up 5 from teaching lab) and 5 kits to go with them BK Precision function generators and Tenma oscilloscopes

More information

Introduction to Machine Learning and Data Mining. Prof. Dr. Igor Trajkovski trajkovski@nyus.edu.mk

Introduction to Machine Learning and Data Mining. Prof. Dr. Igor Trajkovski trajkovski@nyus.edu.mk Introduction to Machine Learning and Data Mining Prof. Dr. Igor Trakovski trakovski@nyus.edu.mk Neural Networks 2 Neural Networks Analogy to biological neural systems, the most robust learning systems

More information

Music Mood Classification

Music Mood Classification Music Mood Classification CS 229 Project Report Jose Padial Ashish Goel Introduction The aim of the project was to develop a music mood classifier. There are many categories of mood into which songs may

More information

High-Performance Signature Recognition Method using SVM

High-Performance Signature Recognition Method using SVM High-Performance Signature Recognition Method using SVM Saeid Fazli Research Institute of Modern Biological Techniques University of Zanjan Shima Pouyan Electrical Engineering Department University of

More information

Analysis/resynthesis with the short time Fourier transform

Analysis/resynthesis with the short time Fourier transform Analysis/resynthesis with the short time Fourier transform summer 2006 lecture on analysis, modeling and transformation of audio signals Axel Röbel Institute of communication science TU-Berlin IRCAM Analysis/Synthesis

More information

1 Topic. 2 Scilab. 2.1 What is Scilab?

1 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 information

Supporting Online Material for

Supporting Online Material for www.sciencemag.org/cgi/content/full/313/5786/504/dc1 Supporting Online Material for Reducing the Dimensionality of Data with Neural Networks G. E. Hinton* and R. R. Salakhutdinov *To whom correspondence

More information

Trading Strategies and the Cat Tournament Protocol

Trading 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 information

Sub-pixel mapping: A comparison of techniques

Sub-pixel mapping: A comparison of techniques Sub-pixel mapping: A comparison of techniques Koen C. Mertens, Lieven P.C. Verbeke & Robert R. De Wulf Laboratory of Forest Management and Spatial Information Techniques, Ghent University, 9000 Gent, Belgium

More information

Detecting Network Anomalies. Anant Shah

Detecting Network Anomalies. Anant Shah Detecting Network Anomalies using Traffic Modeling Anant Shah Anomaly Detection Anomalies are deviations from established behavior In most cases anomalies are indications of problems The science of extracting

More information

PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 4: LINEAR MODELS FOR CLASSIFICATION

PATTERN 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 information

Unconstrained Handwritten Character Recognition Using Different Classification Strategies

Unconstrained Handwritten Character Recognition Using Different Classification Strategies Unconstrained Handwritten Character Recognition Using Different Classification Strategies Alessandro L. Koerich Department of Computer Science (PPGIA) Pontifical Catholic University of Paraná (PUCPR) Curitiba,

More information

Colour Image Segmentation Technique for Screen Printing

Colour Image Segmentation Technique for Screen Printing 60 R.U. Hewage and D.U.J. Sonnadara Department of Physics, University of Colombo, Sri Lanka ABSTRACT Screen-printing is an industry with a large number of applications ranging from printing mobile phone

More information

Lecture 2: The SVM classifier

Lecture 2: The SVM classifier Lecture 2: The SVM classifier C19 Machine Learning Hilary 2015 A. Zisserman Review of linear classifiers Linear separability Perceptron Support Vector Machine (SVM) classifier Wide margin Cost function

More information

School Class Monitoring System Based on Audio Signal Processing

School Class Monitoring System Based on Audio Signal Processing C. R. Rashmi 1,,C.P.Shantala 2 andt.r.yashavanth 3 1 Department of CSE, PG Student, CIT, Gubbi, Tumkur, Karnataka, India. 2 Department of CSE, Vice Principal & HOD, CIT, Gubbi, Tumkur, Karnataka, India.

More information

A Content based Spam Filtering Using Optical Back Propagation Technique

A 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 information

Audio Content Analysis for Online Audiovisual Data Segmentation and Classification

Audio Content Analysis for Online Audiovisual Data Segmentation and Classification IEEE TRANSACTIONS ON SPEECH AND AUDIO PROCESSING, VOL. 9, NO. 4, MAY 2001 441 Audio Content Analysis for Online Audiovisual Data Segmentation and Classification Tong Zhang, Member, IEEE, and C.-C. Jay

More information

SUCCESSFUL PREDICTION OF HORSE RACING RESULTS USING A NEURAL NETWORK

SUCCESSFUL PREDICTION OF HORSE RACING RESULTS USING A NEURAL NETWORK SUCCESSFUL PREDICTION OF HORSE RACING RESULTS USING A NEURAL NETWORK N M Allinson and D Merritt 1 Introduction This contribution has two main sections. The first discusses some aspects of multilayer perceptrons,

More information

Mining of an electrocardiogram

Mining of an electrocardiogram Annales UMCS Informatica AI 4 (2006) 218-229 Annales UMCS Informatica Lublin-Polonia Sectio AI http://www.annales.umcs.lublin.pl/ Mining of an electrocardiogram Urszula Markowska-Kaczmar *, Bartosz Kordas

More information

Lecture 6. Artificial Neural Networks

Lecture 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 information

Lab 1. The Fourier Transform

Lab 1. The Fourier Transform Lab 1. The Fourier Transform Introduction In the Communication Labs you will be given the opportunity to apply the theory learned in Communication Systems. Since this is your first time to work in the

More information

INTELLIGENT ENERGY MANAGEMENT OF ELECTRICAL POWER SYSTEMS WITH DISTRIBUTED FEEDING ON THE BASIS OF FORECASTS OF DEMAND AND GENERATION Chr.

INTELLIGENT ENERGY MANAGEMENT OF ELECTRICAL POWER SYSTEMS WITH DISTRIBUTED FEEDING ON THE BASIS OF FORECASTS OF DEMAND AND GENERATION Chr. INTELLIGENT ENERGY MANAGEMENT OF ELECTRICAL POWER SYSTEMS WITH DISTRIBUTED FEEDING ON THE BASIS OF FORECASTS OF DEMAND AND GENERATION Chr. Meisenbach M. Hable G. Winkler P. Meier Technology, Laboratory

More information

Non-negative Matrix Factorization (NMF) in Semi-supervised Learning Reducing Dimension and Maintaining Meaning

Non-negative Matrix Factorization (NMF) in Semi-supervised Learning Reducing Dimension and Maintaining Meaning Non-negative Matrix Factorization (NMF) in Semi-supervised Learning Reducing Dimension and Maintaining Meaning SAMSI 10 May 2013 Outline Introduction to NMF Applications Motivations NMF as a middle step

More information

Follow links Class Use and other Permissions. For more information, send email to: permissions@pupress.princeton.edu

Follow links Class Use and other Permissions. For more information, send email to: permissions@pupress.princeton.edu COPYRIGHT NOTICE: David A. Kendrick, P. Ruben Mercado, and Hans M. Amman: Computational Economics is published by Princeton University Press and copyrighted, 2006, by Princeton University Press. All rights

More information

EMOTION IN SPEECH: RECOGNITION AND APPLICATION TO CALL CENTERS

EMOTION IN SPEECH: RECOGNITION AND APPLICATION TO CALL CENTERS EMOTION IN SPEECH: RECOGNITION AND APPLICATION TO CALL CENTERS VALERY A. PETRUSHIN Andersen Consulting 3773 Willow Rd. Northbrook, IL 60062 petr@cstar.ac.com ABSTRACT The paper describes two experimental

More information

New Work Item for ISO 3534-5 Predictive Analytics (Initial Notes and Thoughts) Introduction

New 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 information

The Scientific Data Mining Process

The 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 information

Tracking Moving Objects In Video Sequences Yiwei Wang, Robert E. Van Dyck, and John F. Doherty Department of Electrical Engineering The Pennsylvania State University University Park, PA16802 Abstract{Object

More information

SEARCH AND CLASSIFICATION OF "INTERESTING" BUSINESS APPLICATIONS IN THE WORLD WIDE WEB USING A NEURAL NETWORK APPROACH

SEARCH AND CLASSIFICATION OF INTERESTING BUSINESS APPLICATIONS IN THE WORLD WIDE WEB USING A NEURAL NETWORK APPROACH SEARCH AND CLASSIFICATION OF "INTERESTING" BUSINESS APPLICATIONS IN THE WORLD WIDE WEB USING A NEURAL NETWORK APPROACH Abstract Karl Kurbel, Kirti Singh, Frank Teuteberg Europe University Viadrina Frankfurt

More information

Direct and Reflected: Understanding the Truth with Y-S 3

Direct and Reflected: Understanding the Truth with Y-S 3 Direct and Reflected: Understanding the Truth with Y-S 3 -Speaker System Design Guide- December 2008 2008 Yamaha Corporation 1 Introduction Y-S 3 is a speaker system design software application. It is

More information

Developing an Isolated Word Recognition System in MATLAB

Developing an Isolated Word Recognition System in MATLAB MATLAB Digest Developing an Isolated Word Recognition System in MATLAB By Daryl Ning Speech-recognition technology is embedded in voice-activated routing systems at customer call centres, voice dialling

More information

Statistical Machine Learning

Statistical Machine Learning Statistical Machine Learning UoC Stats 37700, Winter quarter Lecture 4: classical linear and quadratic discriminants. 1 / 25 Linear separation For two classes in R d : simple idea: separate the classes

More information

Analog and Digital Signals, Time and Frequency Representation of Signals

Analog and Digital Signals, Time and Frequency Representation of Signals 1 Analog and Digital Signals, Time and Frequency Representation of Signals Required reading: Garcia 3.1, 3.2 CSE 3213, Fall 2010 Instructor: N. Vlajic 2 Data vs. Signal Analog vs. Digital Analog Signals

More information

An Introduction to Data Mining

An 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 information

Method of Combining the Degrees of Similarity in Handwritten Signature Authentication Using Neural Networks

Method of Combining the Degrees of Similarity in Handwritten Signature Authentication Using Neural Networks Method of Combining the Degrees of Similarity in Handwritten Signature Authentication Using Neural Networks Ph. D. Student, Eng. Eusebiu Marcu Abstract This paper introduces a new method of combining the

More information

ELLIOTT WAVES RECOGNITION VIA NEURAL NETWORKS

ELLIOTT WAVES RECOGNITION VIA NEURAL NETWORKS ELLIOTT WAVES RECOGNITION VIA NEURAL NETWORKS Martin Kotyrba Eva Volna David Brazina Robert Jarusek Department of Informatics and Computers University of Ostrava Z70103, Ostrava, Czech Republic martin.kotyrba@osu.cz

More information

Modelling, Extraction and Description of Intrinsic Cues of High Resolution Satellite Images: Independent Component Analysis based approaches

Modelling, Extraction and Description of Intrinsic Cues of High Resolution Satellite Images: Independent Component Analysis based approaches Modelling, Extraction and Description of Intrinsic Cues of High Resolution Satellite Images: Independent Component Analysis based approaches PhD Thesis by Payam Birjandi Director: Prof. Mihai Datcu Problematic

More information

Environmental Remote Sensing GEOG 2021

Environmental Remote Sensing GEOG 2021 Environmental Remote Sensing GEOG 2021 Lecture 4 Image classification 2 Purpose categorising data data abstraction / simplification data interpretation mapping for land cover mapping use land cover class

More information

Automatic Algorithm Design by Neural Network and Indicators in Iran Stock Transactions

Automatic Algorithm Design by Neural Network and Indicators in Iran Stock Transactions Journal of Novel Applied Sciences Available online at www.jnasci.org 2015 JNAS Journal-2015-4-4/501-507 ISSN 2322-5149 2015 JNAS Automatic Algorithm Design by Neural Network and Indicators in Iran Stock

More information

The Fourier Analysis Tool in Microsoft Excel

The Fourier Analysis Tool in Microsoft Excel The Fourier Analysis Tool in Microsoft Excel Douglas A. Kerr Issue March 4, 2009 ABSTRACT AD ITRODUCTIO The spreadsheet application Microsoft Excel includes a tool that will calculate the discrete Fourier

More information