Download presentation
Presentation is loading. Please wait.
1
S. Mandayam/ ANN/ECE Dept./Rowan University Artificial Neural Networks ECE.09.454/ECE.09.560 Fall 2010 Shreekanth Mandayam ECE Department Rowan University http://engineering.rowan.edu/~shreek/fall10/ann/ Lecture 2 September 20, 2010
2
S. Mandayam/ ANN/ECE Dept./Rowan UniversityPlan Recall: Neural Network Paradigm Recall: Perceptron Model Learning Processes Rules Paradigms Tasks Perceptron Training Algorithm Widrow-Hoff Rule (LMS Algorithm) Lab Project 1
3
S. Mandayam/ ANN/ECE Dept./Rowan University Recall: Neural Network Paradigm Stage 1: Network Training ArtificialNeuralNetwork Present Examples Indicate Desired Outputs DetermineSynapticWeights ArtificialNeuralNetwork New Data Predicted Outputs Stage 2: Network Testing “knowledge”
4
S. Mandayam/ ANN/ECE Dept./Rowan University Recall: ANN Model ArtificialNeuralNetwork x Input Vector y Output Vector f Complex Nonlinear Function f(x) = y “knowledge”
5
S. Mandayam/ ANN/ECE Dept./Rowan University Recall: The Perceptron Model (.) w k1 w k2 w km x1x1 x2x2 xmxm Inputs Synaptic weights Bias, b k Induced field, v k Output, y k ukuk Activation/ squashing function
6
S. Mandayam/ ANN/ECE Dept./Rowan University“Learning” [w] x y ANN Mathematical Model of the Learning Process [w] 0 x y(0) Intitialize: Iteration (0) [w] 1 x y(1) Iteration (1) [w] n x y(n) = d Iteration (n) desired o/p
7
S. Mandayam/ ANN/ECE Dept./Rowan University Learning Rules Error Correction Learning Delta Rule or Widrow-Hoff Rule Memory Based Learning Nearest Neighbor Rule Hebbian Learning Competitive Learning Boltzman Learning
8
S. Mandayam/ ANN/ECE Dept./Rowan University Error-Correction Learning (.) w k1 (n) x 1 (n) x2x2 xmxm Inputs Synaptic weights Bias, b k Induced field, v k (n) Activation/ squashing function w k2 (n) w km (n) Output, y k (n) Desired Output, d k (n) Error Signal e k (n) + -
9
S. Mandayam/ ANN/ECE Dept./Rowan University Learning Paradigms Environment (Data) Teacher (Expert) ANN error desired actual + - Supervised Unsupervised
10
S. Mandayam/ ANN/ECE Dept./Rowan University Learning Paradigms Supervised Unsupervised Environment (Data) Delay ANN Delayed Reinforcement Learning Cost Function
11
S. Mandayam/ ANN/ECE Dept./Rowan University Learning Tasks Pattern Association Pattern Recognition Function Approximation Filtering Classification x1x1 x2x2 1 2 DB x1x1 x2x2 1 2
12
S. Mandayam/ ANN/ECE Dept./Rowan University Perceptron Training Widrow-Hoff Rule (LMS Algorithm) w(0) = 0 n = 0 y(n) = sgn [w T (n) x(n)] w(n+1) = w(n) + [d(n) – y(n)]x(n) n = n+1 Matlab Demo
13
S. Mandayam/ ANN/ECE Dept./Rowan University Lab Project 1 http://engineering.rowan.edu/~shreek /fall10/ann/lab1.htmlhttp://engineering.rowan.edu/~shreek /fall10/ann/lab1.html
14
S. Mandayam/ ANN/ECE Dept./Rowan University Lab Project 1 http://engineering.rowan.edu/~shreek/fall10/ann/lab1.html Double-moon Classification Problem
15
S. Mandayam/ ANN/ECE Dept./Rowan University Lab Project 1 http://engineering.rowan.edu/~shreek/fall10/ann/lab1.html Double-moon Classification Problem d = 1; linearly separable d = -4; NOT linearly separable
16
S. Mandayam/ ANN/ECE Dept./Rowan University Lab Project 1 http://engineering.rowan.edu/~shreek /fall10/ann/lab1.htmlhttp://engineering.rowan.edu/~shreek /fall10/ann/lab1.html UCI Machine Learning Repository: http://www.ics.uci.edu/~mlearn/MLRepository.html Face Recognition: Generate images
17
S. Mandayam/ ANN/ECE Dept./Rowan UniversitySummary
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.