SlideShare a Scribd company logo
1 of 16
Simon de Montignyand Richard Labib ÉcolePolytechnique de Montréal Département de Mathématiques et GénieIndustriel Learning Algorithms for a Specific Configuration of the Quantron
Introduction The Quantron is a new neuron model solves nonlinear problems no efficient learning algorithm Goal : retrieve parameters from output image For a specific configuration using surrogate potential functions, we can express the activation function analytically parameterize the decision boundary develop convergent learning algorithms
Outline Review of the Quantron Surrogate potential functions Decision boundary and image analysis Learning algorithms and results Summary
Review of the Quantron Hybrid neuron model Spatial and temporal summation of potentials Input value = delay between potentials Fires when sum of potentials reaches threshold Sum of potentials Threshold Neuron fires
Review of the Quantron Activation function Use in classification Output is 1 if max𝐴(𝑡)≥Γ and 0 otherwise   shift parameter inputs 𝐴𝑡=𝑖=0𝑁1−1𝑝1(𝑡−𝜃1−𝑖𝑥)+𝑗=0𝑁2−1𝑝2(𝑡−𝜃2−𝑗𝑦)   shift parameter potentials
Surrogate potential functions We use rectangular and ramp potentials.
Surrogate potential functions Specific configuration of the Quantron two inputs : 𝑥>0 and 𝑦>0 weights : 𝑤1≥0 and 𝑤2≥0 widths : 𝑟1≥0and 𝑟2≥0 shift parameters 𝜃1 and 𝜃2 used to synchronize the end time of the first potential from each input infinite number of potentials We obtain analytical expressions for max𝐴(𝑡) involving ceiling functions.   
Surrogate potential functions Rectangular potentials max𝐴(𝑡)=𝑤1𝑟1𝑥3&<𝑦>+𝑤2𝑟2𝑦   Ramp potentials max𝐴(𝑡)=𝑤1𝑟1𝑥3&<𝑦>1−𝑥2𝑟1𝑟1𝑥3&<𝑦>−1                        +𝑤2𝑟2𝑦1−𝑦2𝑟2𝑟2𝑦−1  
Decision boundary and image analysis Rectangular potentials Corner coordinates (𝑎,𝑏) and 𝑐,𝑑 are linked to 𝑤1,𝑤2,𝑟1,𝑟2 by a non-invertible equation system.   On a pixel grid, corners are located inside a square found by analyzing pixel rows and columns.
Decision boundary and image analysis Ramp potentials Corner coordinates (𝑎,𝑏) and 𝑐,𝑑 are linked to 𝑤1,𝑤2,𝑟1,𝑟2 by an invertible equation system.   On a pixel grid, corners are located inside a polygon provided by a custom image analysis algorithm.
Learning algorithm (rect.) With rectangular potentials, if 𝑤1≤𝑤2 : we have 𝑟2=𝑏 there is an integer 𝑚 for which 𝑟1=𝑎𝑚 If 𝑤1≥𝑤2 : we have 𝑟1=𝑐  there is an integer 𝑚 for which𝑟2=𝑑𝑚 𝑚≥ total number of corners in boundary For a fixed value of 𝑚, we select corner coordinates randomly in squares and set the values of 𝑟1 and 𝑟2.  
Learning algorithm (rect.) To train the Quantron efficiently : we consider both 𝑤1≤𝑤2 and 𝑤1≥𝑤2; we minimize sum-of-squares error functions sequentially for different values of 𝑚; we stop if zero misclassification rate is reached. Convergence of the algorithm max𝐴(𝑡) is linear in 𝑤1 and 𝑤2 unimodal error function  
Results on test problem (rect.) 𝑤1≤𝑤2𝑤1≥𝑤2  
Learning algorithm (ramp) Using ramp potentials, we can train the Quantron in a single step by inverting the system linking the corner coordinates and the parameters. We choose corner coordinates randomly in polygons and obtain parameter values. We repeat this procedure to obtain a solution with a low misclassification rate.
Results on test problem (ramp) Misclassification rate for 50 random trials
Summary We obtained convergent learning algorithms for a specific configuration of the Quantron. Rect. : sequence of unimodal error functions Ramp : analytical solution to system of equations These algorithms depend on precise geometric characteristics. Future research : generalization to real classification data

More Related Content

What's hot

WolframAlpha Examples part 4
WolframAlpha Examples part 4WolframAlpha Examples part 4
WolframAlpha Examples part 4Colleen Young
 
MATLAB/SIMULINK for engineering applications: day 3
MATLAB/SIMULINK for engineering applications: day 3MATLAB/SIMULINK for engineering applications: day 3
MATLAB/SIMULINK for engineering applications: day 3reddyprasad reddyvari
 
Polymath For Chemical Engineers
Polymath For Chemical EngineersPolymath For Chemical Engineers
Polymath For Chemical EngineersHashim Khan
 
Modeling of Wireless Power Transfer by COMSOL: A Quick Tutorial
Modeling of Wireless Power Transfer by COMSOL: A Quick TutorialModeling of Wireless Power Transfer by COMSOL: A Quick Tutorial
Modeling of Wireless Power Transfer by COMSOL: A Quick TutorialAmirhossein Hajiaghajani
 
Matlab for Electrical Engineers
Matlab for Electrical EngineersMatlab for Electrical Engineers
Matlab for Electrical EngineersManish Joshi
 
Introduction to simulink (1)
Introduction to simulink (1)Introduction to simulink (1)
Introduction to simulink (1)Memo Love
 
Introduction to MATLAB 1
Introduction to MATLAB 1Introduction to MATLAB 1
Introduction to MATLAB 1Mohamed Gafar
 
Why you should use a testing framework
Why you should use a testing frameworkWhy you should use a testing framework
Why you should use a testing frameworkRichie Cotton
 
CS106 Lab 4 - If statement
CS106 Lab 4 - If statementCS106 Lab 4 - If statement
CS106 Lab 4 - If statementNada Kamel
 
Bubble Sort algorithm in Assembly Language
Bubble Sort algorithm in Assembly LanguageBubble Sort algorithm in Assembly Language
Bubble Sort algorithm in Assembly LanguageAriel Tonatiuh Espindola
 
mc_simulation documentation
mc_simulation documentationmc_simulation documentation
mc_simulation documentationCarlo Parodi
 
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMSNEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMSESCOM
 
Forelasning4
Forelasning4Forelasning4
Forelasning4Memo Love
 
Introduction to Matlab
Introduction to MatlabIntroduction to Matlab
Introduction to MatlabAmr Rashed
 
Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...
Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...
Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...infopapers
 

What's hot (20)

WolframAlpha Examples part 4
WolframAlpha Examples part 4WolframAlpha Examples part 4
WolframAlpha Examples part 4
 
MATLAB/SIMULINK for engineering applications: day 3
MATLAB/SIMULINK for engineering applications: day 3MATLAB/SIMULINK for engineering applications: day 3
MATLAB/SIMULINK for engineering applications: day 3
 
Polymath For Chemical Engineers
Polymath For Chemical EngineersPolymath For Chemical Engineers
Polymath For Chemical Engineers
 
#1 designandanalysis of algo
#1 designandanalysis of algo#1 designandanalysis of algo
#1 designandanalysis of algo
 
Modeling of Wireless Power Transfer by COMSOL: A Quick Tutorial
Modeling of Wireless Power Transfer by COMSOL: A Quick TutorialModeling of Wireless Power Transfer by COMSOL: A Quick Tutorial
Modeling of Wireless Power Transfer by COMSOL: A Quick Tutorial
 
Matlab for Electrical Engineers
Matlab for Electrical EngineersMatlab for Electrical Engineers
Matlab for Electrical Engineers
 
Nonnegative Matrix Factorization with Side Information for Time Series Recove...
Nonnegative Matrix Factorization with Side Information for Time Series Recove...Nonnegative Matrix Factorization with Side Information for Time Series Recove...
Nonnegative Matrix Factorization with Side Information for Time Series Recove...
 
Introduction to simulink (1)
Introduction to simulink (1)Introduction to simulink (1)
Introduction to simulink (1)
 
Introduction to MATLAB 1
Introduction to MATLAB 1Introduction to MATLAB 1
Introduction to MATLAB 1
 
All projects
All projectsAll projects
All projects
 
ANCLMS
ANCLMSANCLMS
ANCLMS
 
Why you should use a testing framework
Why you should use a testing frameworkWhy you should use a testing framework
Why you should use a testing framework
 
CS106 Lab 4 - If statement
CS106 Lab 4 - If statementCS106 Lab 4 - If statement
CS106 Lab 4 - If statement
 
DSP lab manual
DSP lab manualDSP lab manual
DSP lab manual
 
Bubble Sort algorithm in Assembly Language
Bubble Sort algorithm in Assembly LanguageBubble Sort algorithm in Assembly Language
Bubble Sort algorithm in Assembly Language
 
mc_simulation documentation
mc_simulation documentationmc_simulation documentation
mc_simulation documentation
 
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMSNEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS
NEURAL NETWORK Widrow-Hoff Learning Adaline Hagan LMS
 
Forelasning4
Forelasning4Forelasning4
Forelasning4
 
Introduction to Matlab
Introduction to MatlabIntroduction to Matlab
Introduction to Matlab
 
Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...
Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...
Automatic control based on Wasp Behavioral Model and Stochastic Learning Auto...
 

Similar to Learning Algorithms For A Specific Configuration Of The Quantron

SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15Hao Zhuang
 
05 contours seg_matching
05 contours seg_matching05 contours seg_matching
05 contours seg_matchingankit_ppt
 
Kulum alin-11 jan2014
Kulum alin-11 jan2014Kulum alin-11 jan2014
Kulum alin-11 jan2014rolly purnomo
 
Efficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketchingEfficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketchingHsing-chuan Hsieh
 
Learning machine learning with Yellowbrick
Learning machine learning with YellowbrickLearning machine learning with Yellowbrick
Learning machine learning with YellowbrickRebecca Bilbro
 
Distributed_Array_Algos.pptx
Distributed_Array_Algos.pptxDistributed_Array_Algos.pptx
Distributed_Array_Algos.pptxf20170649g
 
Dmitrii Tihonkih - The Iterative Closest Points Algorithm and Affine Transfo...
Dmitrii Tihonkih - The Iterative Closest Points Algorithm and  Affine Transfo...Dmitrii Tihonkih - The Iterative Closest Points Algorithm and  Affine Transfo...
Dmitrii Tihonkih - The Iterative Closest Points Algorithm and Affine Transfo...AIST
 
Beginners Guide to Non-Negative Matrix Factorization
Beginners Guide to Non-Negative Matrix FactorizationBeginners Guide to Non-Negative Matrix Factorization
Beginners Guide to Non-Negative Matrix FactorizationBenjamin Bengfort
 
QTML2021 UAP Quantum Feature Map
QTML2021 UAP Quantum Feature MapQTML2021 UAP Quantum Feature Map
QTML2021 UAP Quantum Feature MapHa Phuong
 
EE660_Report_YaxinLiu_8448347171
EE660_Report_YaxinLiu_8448347171EE660_Report_YaxinLiu_8448347171
EE660_Report_YaxinLiu_8448347171Yaxin Liu
 
Randomized Algorithm- Advanced Algorithm
Randomized Algorithm- Advanced AlgorithmRandomized Algorithm- Advanced Algorithm
Randomized Algorithm- Advanced AlgorithmMahbubur Rahman
 
Numerical Methods
Numerical MethodsNumerical Methods
Numerical MethodsESUG
 
Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)
Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)
Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)ALINLAB
 
Linear regression, costs & gradient descent
Linear regression, costs & gradient descentLinear regression, costs & gradient descent
Linear regression, costs & gradient descentRevanth Kumar
 
Forecasting Default Probabilities in Emerging Markets and Dynamical Regula...
Forecasting Default Probabilities  in Emerging Markets and   Dynamical Regula...Forecasting Default Probabilities  in Emerging Markets and   Dynamical Regula...
Forecasting Default Probabilities in Emerging Markets and Dynamical Regula...SSA KPI
 
CS345-Algorithms-II-Lecture-1-CS345-2016.pdf
CS345-Algorithms-II-Lecture-1-CS345-2016.pdfCS345-Algorithms-II-Lecture-1-CS345-2016.pdf
CS345-Algorithms-II-Lecture-1-CS345-2016.pdfOpenWorld6
 
Machine Learning.pdf
Machine Learning.pdfMachine Learning.pdf
Machine Learning.pdfBeyaNasr1
 

Similar to Learning Algorithms For A Specific Configuration Of The Quantron (20)

SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15SPICE-MATEX @ DAC15
SPICE-MATEX @ DAC15
 
05 contours seg_matching
05 contours seg_matching05 contours seg_matching
05 contours seg_matching
 
Kulum alin-11 jan2014
Kulum alin-11 jan2014Kulum alin-11 jan2014
Kulum alin-11 jan2014
 
Efficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketchingEfficient anomaly detection via matrix sketching
Efficient anomaly detection via matrix sketching
 
Learning machine learning with Yellowbrick
Learning machine learning with YellowbrickLearning machine learning with Yellowbrick
Learning machine learning with Yellowbrick
 
Distributed_Array_Algos.pptx
Distributed_Array_Algos.pptxDistributed_Array_Algos.pptx
Distributed_Array_Algos.pptx
 
Dmitrii Tihonkih - The Iterative Closest Points Algorithm and Affine Transfo...
Dmitrii Tihonkih - The Iterative Closest Points Algorithm and  Affine Transfo...Dmitrii Tihonkih - The Iterative Closest Points Algorithm and  Affine Transfo...
Dmitrii Tihonkih - The Iterative Closest Points Algorithm and Affine Transfo...
 
Beginners Guide to Non-Negative Matrix Factorization
Beginners Guide to Non-Negative Matrix FactorizationBeginners Guide to Non-Negative Matrix Factorization
Beginners Guide to Non-Negative Matrix Factorization
 
QTML2021 UAP Quantum Feature Map
QTML2021 UAP Quantum Feature MapQTML2021 UAP Quantum Feature Map
QTML2021 UAP Quantum Feature Map
 
EE660_Report_YaxinLiu_8448347171
EE660_Report_YaxinLiu_8448347171EE660_Report_YaxinLiu_8448347171
EE660_Report_YaxinLiu_8448347171
 
MS Thesis
MS ThesisMS Thesis
MS Thesis
 
MS Thesis
MS ThesisMS Thesis
MS Thesis
 
Randomized Algorithm- Advanced Algorithm
Randomized Algorithm- Advanced AlgorithmRandomized Algorithm- Advanced Algorithm
Randomized Algorithm- Advanced Algorithm
 
Numerical Methods
Numerical MethodsNumerical Methods
Numerical Methods
 
Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)
Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)
Polynomial Tensor Sketch for Element-wise Matrix Function (ICML 2020)
 
Linear regression, costs & gradient descent
Linear regression, costs & gradient descentLinear regression, costs & gradient descent
Linear regression, costs & gradient descent
 
Py data19 final
Py data19   finalPy data19   final
Py data19 final
 
Forecasting Default Probabilities in Emerging Markets and Dynamical Regula...
Forecasting Default Probabilities  in Emerging Markets and   Dynamical Regula...Forecasting Default Probabilities  in Emerging Markets and   Dynamical Regula...
Forecasting Default Probabilities in Emerging Markets and Dynamical Regula...
 
CS345-Algorithms-II-Lecture-1-CS345-2016.pdf
CS345-Algorithms-II-Lecture-1-CS345-2016.pdfCS345-Algorithms-II-Lecture-1-CS345-2016.pdf
CS345-Algorithms-II-Lecture-1-CS345-2016.pdf
 
Machine Learning.pdf
Machine Learning.pdfMachine Learning.pdf
Machine Learning.pdf
 

Learning Algorithms For A Specific Configuration Of The Quantron

  • 1. Simon de Montignyand Richard Labib ÉcolePolytechnique de Montréal Département de Mathématiques et GénieIndustriel Learning Algorithms for a Specific Configuration of the Quantron
  • 2. Introduction The Quantron is a new neuron model solves nonlinear problems no efficient learning algorithm Goal : retrieve parameters from output image For a specific configuration using surrogate potential functions, we can express the activation function analytically parameterize the decision boundary develop convergent learning algorithms
  • 3. Outline Review of the Quantron Surrogate potential functions Decision boundary and image analysis Learning algorithms and results Summary
  • 4. Review of the Quantron Hybrid neuron model Spatial and temporal summation of potentials Input value = delay between potentials Fires when sum of potentials reaches threshold Sum of potentials Threshold Neuron fires
  • 5. Review of the Quantron Activation function Use in classification Output is 1 if max𝐴(𝑡)≥Γ and 0 otherwise   shift parameter inputs 𝐴𝑡=𝑖=0𝑁1−1𝑝1(𝑡−𝜃1−𝑖𝑥)+𝑗=0𝑁2−1𝑝2(𝑡−𝜃2−𝑗𝑦)   shift parameter potentials
  • 6. Surrogate potential functions We use rectangular and ramp potentials.
  • 7. Surrogate potential functions Specific configuration of the Quantron two inputs : 𝑥>0 and 𝑦>0 weights : 𝑤1≥0 and 𝑤2≥0 widths : 𝑟1≥0and 𝑟2≥0 shift parameters 𝜃1 and 𝜃2 used to synchronize the end time of the first potential from each input infinite number of potentials We obtain analytical expressions for max𝐴(𝑡) involving ceiling functions.  
  • 8. Surrogate potential functions Rectangular potentials max𝐴(𝑡)=𝑤1𝑟1𝑥3&<𝑦>+𝑤2𝑟2𝑦   Ramp potentials max𝐴(𝑡)=𝑤1𝑟1𝑥3&<𝑦>1−𝑥2𝑟1𝑟1𝑥3&<𝑦>−1                        +𝑤2𝑟2𝑦1−𝑦2𝑟2𝑟2𝑦−1  
  • 9. Decision boundary and image analysis Rectangular potentials Corner coordinates (𝑎,𝑏) and 𝑐,𝑑 are linked to 𝑤1,𝑤2,𝑟1,𝑟2 by a non-invertible equation system.   On a pixel grid, corners are located inside a square found by analyzing pixel rows and columns.
  • 10. Decision boundary and image analysis Ramp potentials Corner coordinates (𝑎,𝑏) and 𝑐,𝑑 are linked to 𝑤1,𝑤2,𝑟1,𝑟2 by an invertible equation system.   On a pixel grid, corners are located inside a polygon provided by a custom image analysis algorithm.
  • 11. Learning algorithm (rect.) With rectangular potentials, if 𝑤1≤𝑤2 : we have 𝑟2=𝑏 there is an integer 𝑚 for which 𝑟1=𝑎𝑚 If 𝑤1≥𝑤2 : we have 𝑟1=𝑐  there is an integer 𝑚 for which𝑟2=𝑑𝑚 𝑚≥ total number of corners in boundary For a fixed value of 𝑚, we select corner coordinates randomly in squares and set the values of 𝑟1 and 𝑟2.  
  • 12. Learning algorithm (rect.) To train the Quantron efficiently : we consider both 𝑤1≤𝑤2 and 𝑤1≥𝑤2; we minimize sum-of-squares error functions sequentially for different values of 𝑚; we stop if zero misclassification rate is reached. Convergence of the algorithm max𝐴(𝑡) is linear in 𝑤1 and 𝑤2 unimodal error function  
  • 13. Results on test problem (rect.) 𝑤1≤𝑤2𝑤1≥𝑤2  
  • 14. Learning algorithm (ramp) Using ramp potentials, we can train the Quantron in a single step by inverting the system linking the corner coordinates and the parameters. We choose corner coordinates randomly in polygons and obtain parameter values. We repeat this procedure to obtain a solution with a low misclassification rate.
  • 15. Results on test problem (ramp) Misclassification rate for 50 random trials
  • 16. Summary We obtained convergent learning algorithms for a specific configuration of the Quantron. Rect. : sequence of unimodal error functions Ramp : analytical solution to system of equations These algorithms depend on precise geometric characteristics. Future research : generalization to real classification data