Connectionist models. Connectionist Models Motivated by Brain rather than Mind –A large number of very simple processing elements –A large number of weighted.

Slides:



Advertisements
Similar presentations
Artificial Neural Networks
Advertisements

Multi-Layer Perceptron (MLP)
1 Machine Learning: Lecture 4 Artificial Neural Networks (Based on Chapter 4 of Mitchell T.., Machine Learning, 1997)
Artificial Neural Networks (1)
Neural Networks  A neural network is a network of simulated neurons that can be used to recognize instances of patterns. NNs learn by searching through.
Machine Learning: Connectionist McCulloch-Pitts Neuron Perceptrons Multilayer Networks Support Vector Machines Feedback Networks Hopfield Networks.
Classification Neural Networks 1
Machine Learning Neural Networks
Simple Neural Nets For Pattern Classification
Neural Networks Basic concepts ArchitectureOperation.
1 Chapter 11 Neural Networks. 2 Chapter 11 Contents (1) l Biological Neurons l Artificial Neurons l Perceptrons l Multilayer Neural Networks l Backpropagation.
Rutgers CS440, Fall 2003 Neural networks Reading: Ch. 20, Sec. 5, AIMA 2 nd Ed.
Connectionist Modeling Some material taken from cspeech.ucd.ie/~connectionism and Rich & Knight, 1991.
Neural Networks. R & G Chapter Feed-Forward Neural Networks otherwise known as The Multi-layer Perceptron or The Back-Propagation Neural Network.
An Illustrative Example
Chapter Seven The Network Approach: Mind as a Web.
Data Mining with Neural Networks (HK: Chapter 7.5)
CS 4700: Foundations of Artificial Intelligence
Neural Networks. Background - Neural Networks can be : Biological - Biological models Artificial - Artificial models - Desire to produce artificial systems.
CSC 4510 – Machine Learning Dr. Mary-Angela Papalaskari Department of Computing Sciences Villanova University Course website:
Artificial Neural Networks
Presentation on Neural Networks.. Basics Of Neural Networks Neural networks refers to a connectionist model that simulates the biophysical information.
Computer Science and Engineering
Artificial Neural Networks
Artificial Neural Nets and AI Connectionism Sub symbolic reasoning.
Multi-Layer Perceptrons Michael J. Watts
Neural Networks Ellen Walker Hiram College. Connectionist Architectures Characterized by (Rich & Knight) –Large number of very simple neuron-like processing.
Chapter 9 Neural Network.
Neural Networks AI – Week 23 Sub-symbolic AI Multi-Layer Neural Networks Lee McCluskey, room 3/10
Machine Learning Chapter 4. Artificial Neural Networks
Machine Learning Dr. Shazzad Hosain Department of EECS North South Universtiy
NEURAL NETWORKS FOR DATA MINING
 Diagram of a Neuron  The Simple Perceptron  Multilayer Neural Network  What is Hidden Layer?  Why do we Need a Hidden Layer?  How do Multilayer.
Artificial Intelligence Methods Neural Networks Lecture 4 Rakesh K. Bissoondeeal Rakesh K. Bissoondeeal.
Artificial Intelligence Techniques Multilayer Perceptrons.
Artificial Neural Networks. The Brain How do brains work? How do human brains differ from that of other animals? Can we base models of artificial intelligence.
1 Chapter 11 Neural Networks. 2 Chapter 11 Contents (1) l Biological Neurons l Artificial Neurons l Perceptrons l Multilayer Neural Networks l Backpropagation.
Neural Networks and Backpropagation Sebastian Thrun , Fall 2000.
Neural Network Basics Anns are analytical systems that address problems whose solutions have not been explicitly formulated Structure in which multiple.
Back-Propagation Algorithm AN INTRODUCTION TO LEARNING INTERNAL REPRESENTATIONS BY ERROR PROPAGATION Presented by: Kunal Parmar UHID:
CS621 : Artificial Intelligence
Neural Networks Presented by M. Abbasi Course lecturer: Dr.Tohidkhah.
Neural Networks Teacher: Elena Marchiori R4.47 Assistant: Kees Jong S2.22
Chapter 18 Connectionist Models
EEE502 Pattern Recognition
COMP53311 Other Classification Models: Neural Network Prepared by Raymond Wong Some of the notes about Neural Network are borrowed from LW Chan’s notes.
Perceptrons Michael J. Watts
Previous Lecture Perceptron W  t+1  W  t  t  d(t) - sign (w(t)  x)] x Adaline W  t+1  W  t  t  d(t) - f(w(t)  x)] f’ x Gradient.
Artificial Intelligence CIS 342 The College of Saint Rose David Goldschmidt, Ph.D.
Chapter 6 Neural Network.
Bab 5 Classification: Alternative Techniques Part 4 Artificial Neural Networks Based Classifer.
“Principles of Soft Computing, 2 nd Edition” by S.N. Sivanandam & SN Deepa Copyright  2011 Wiley India Pvt. Ltd. All rights reserved. CHAPTER 2 ARTIFICIAL.
Data Mining: Concepts and Techniques1 Prediction Prediction vs. classification Classification predicts categorical class label Prediction predicts continuous-valued.
A Presentation on Adaptive Neuro-Fuzzy Inference System using Particle Swarm Optimization and it’s Application By Sumanta Kundu (En.R.No.
CSE343/543 Machine Learning Mayank Vatsa Lecture slides are prepared using several teaching resources and no authorship is claimed for any slides.
Neural Networks: An Introduction and Overview
Neural Network Architecture Session 2
Other Classification Models: Neural Network
Other Classification Models: Neural Network
with Daniel L. Silver, Ph.D. Christian Frey, BBA April 11-12, 2017
CSSE463: Image Recognition Day 17
Classification Neural Networks 1
CSSE463: Image Recognition Day 17
CSSE463: Image Recognition Day 17
CSSE463: Image Recognition Day 17
The Network Approach: Mind as a Web
Introduction to Neural Network
Neural Networks: An Introduction and Overview
David Kauchak CS158 – Spring 2019
Outline Announcement Neural networks Perceptrons - continued
Presentation transcript:

Connectionist models

Connectionist Models Motivated by Brain rather than Mind –A large number of very simple processing elements –A large number of weighted connections between elements (network) –Parallel, distributed control –Emphasis on learning internal representations automatically

The Perceptron w0w0 w1w1 w2w2 w3w3 x1x1 x2x2 x3x3 1 Bias unit Output unit Input units

Training the Perceptron I The original idea here was to train the perceptron by changing the weights in accordance with experience – designed to mimic the real life neuron. If the classification was correct then the weights are unchanged,, however if the classification was wrong then the weights are altered

Training the Perceptron II The weights are altered according to a simple rule: New w i = Old w i + (t-y)x i Here the true outcome is t, the predicted outcome is y, and is the learning rate.

The Method Works! For linearly separable data the method works and will give a line that separates the data into two sets. Even better the method can be proved to work! However for data which is not linearly separable the method does not converge.

Neural Nets The Neural Net is a development from the Perceptron in that several Perceptrons are linked together in a net. Also the threshold function (used in the output unit) is often changed to a continuous function like the logistic function. This leads to a variety of possible architectures.

Multilayered Networks Input layer Hidden layer 1 Hidden layer 2... Hidden layer N Output layer Each layer is fully connected to its preceding and succeeding layers only Every connection has its own weight

Each node at next layer will compute the sigmoid function and propagate values to the next layer Propagate these values forward until output is achieved

Neural Net Inputs Outputs Hidden Layer(s) Bias Units+1 w ij

Neural Net Theory A Neural Net with no hidden layers can classify linearly separable problems A neural net with one hidden layer can describe any continuous function A neural net with two hidden layers can describe any function

Back Propagtion After early interest Neural Nets (NN’s) went into decline as people realized that while you could train perceptrons successfully, this only worked for linearly separable data and no-one had a method to train nets with a hidden layer. The method was suggested by Werbos(1974) but the modern form was given by Rumelhart and McClelland in The technique is based on gradient search techniques.

Backpropagation To train a multilayered network: –randomly initialize all weights [-1..+1] –choose a training example and use feedforward –if correct, backpropagate reward by increasing weights that led to correct output –if incorrect, backpropagate punishment by decreasing weights that led to incorrect output

Backpropagation continued Continue this for each example in training set –This is 1 epoch –After 1 complete epoch, repeat process –Repeat until network has reached a stable state (i.e. changes to weights are always less than some minimum amount that is trivial) Training may take 1000’s or more epochs! (even millions)

Uses of NNs NN are knowledge poor and have internal representations that are meaningless to us However, NN can learn classifications and recognitions Some useful applications include –Pattern recognizers, associative memories, pattern transformers, dynamic transformers

Particular Domains Speech recognition (vowel distinction) Visual Recognition Combinatorial problems Motor-type problems (including vehicular control) Classification-type problems with reasonable sized inputs Game playing (backgammon)

Advantages of NN Able to handle fuzziness Able to handle degraded inputs and ambiguity Able to learn their own internal representations & learn new things Use distributed representations Capable of supervised & unsupervised learning “Easy” to build

Using R to implement Neural Nets R only fits neural nets with one hidden layer There are two ways of fitting neural nets to data in R 1. Via the “model” format NNB.nn2 <- nnet(Type ~ x+y, data = NNeighbour, subset = samp, size = 2, rang = 0.1, decay = 5e-4, maxit = 200)

Using R to implement Neural Nets 2 2.And via the “data” method: NNB.nn1 <- nnet(NNB[samp,], Ntype[samp], size = 2, rang = 0.1, decay = 5e-4, maxit = 200) Note for this method the data needs to be numeric – even the classification data

Using R to implement Neural Nets 3 R uses a random allocation of weights before training the network so each time you perform the calculation you will get different answers It is easy in R to select a sample of the data and to train the data on that sample: Just use the “sample(1:50,25)” function which will select 25 cases at Random from the first 50 cases