Downloaded 324 times
![Supervised Learning
• Child learns from a teacher
• Each input vector requires a corresponding target vector.
• Training pair=[input vector, target vector]
Neural
Network
W
Error
Signal
Generator
X
(Input)
Y
(Actual output)
(Desired Output)
Error
(D-Y)
signals
6Neural Networks Dr. Randa Elanwar](https://image.slidesharecdn.com/lecture7-150201050436-conversion-gate01/75/Introduction-to-Neural-networks-under-graduate-course-Lecture-7-of-9-6-2048.jpg)
![Hebbian learning Rule
• Update weight vector for iteration 1
• Update weight vector for iteration 2
20Neural Networks Dr. Randa Elanwar
11,3
0
5.1
2
1
]5.0011[1.)0(
oXW
5.0
5.1
3
2
1.. 1)0()1( XOWW
TT
12,75.0
5.1
2
5.0
1
]5.05.132[2.)1(
oXW
2
5.3
5.2
1
2.. 2)1()2( XOWW
TT
](https://image.slidesharecdn.com/lecture7-150201050436-conversion-gate01/75/Introduction-to-Neural-networks-under-graduate-course-Lecture-7-of-9-20-2048.jpg)
![Hebbian learning Rule
• Update weight vector for iteration 3
21Neural Networks Dr. Randa Elanwar
13,3]5.1110.[
5.1
1
1
0
]25.35.21[3.)2(
oXW
5.0
5.4
5.3
1
3.. 3)2()3( XOWW
TT
](https://image.slidesharecdn.com/lecture7-150201050436-conversion-gate01/75/Introduction-to-Neural-networks-under-graduate-course-Lecture-7-of-9-21-2048.jpg)
![Hebbian learning Rule
• Update weight vector for iteration 1
• Update weight vector for iteration 2
23Neural Networks Dr. Randa Elanwar
11,0
1
1
1
].000[1.)0(
oXW
1
1
1
1.. 1)0()1( XOWW
TT
12,1
1
1
1
].111[2.)1(
oXW
2
0
2
2.. 2)1()2( XOWW
TT
Ok
Ok](https://image.slidesharecdn.com/lecture7-150201050436-conversion-gate01/75/Introduction-to-Neural-networks-under-graduate-course-Lecture-7-of-9-23-2048.jpg)
![Hebbian learning Rule
• Update weight vector for iteration 3
• Update weight vector for iteration 4
• Since we are given desired outputs we have to correlate them
to the actual outputs i.e. o4 = +1*ydes=-1
24Neural Networks Dr. Randa Elanwar
13,0
1
1
1
].202[3.)2(
oXW
3
1
1
3.. 3)2()3( XOWW
TT
Ok
14,1
1
1
1
].311[4.)3(
oXW
2
2
2
4.. 4)3()4( XOWW
TT
Wrong](https://image.slidesharecdn.com/lecture7-150201050436-conversion-gate01/75/Introduction-to-Neural-networks-under-graduate-course-Lecture-7-of-9-24-2048.jpg)
Uploaded byRanda Elanwar
Introduction to Neural networks (under graduate course) Lecture 7 of 9
This document provides an overview of neural network learning techniques including supervised, unsupervised, and reinforcement learning. It discusses the Hebbian learning rule, which updates weights based on the activation of connected neurons. Examples are provided to illustrate how the Hebbian rule can be used to train networks without error signals by detecting correlations in input-output patterns.
In this document
Powered by AI
Introduction to Neural Networks and mentoring by Dr. Randa Elanwar
Discussion on NN learning techniques: supervised, unsupervised, reinforcement, and Hebbian learning rule.
Overview of the basic models of Artificial Neural Networks (ANN): Activation function, Interconnections, Learning rules.
Explains the adaptation of NN to stimuli through parameter adjustments and the types of learning: Parameter and Structure Learning.
Defines training in neural networks as the process of modifying connection weights using different learning methods.
Describes supervised learning as learning from a teacher with clear input-output pairs, including error signal generation.
Focuses on supervision in learning to minimize error effectively.
Details methods of learning through target inputs, comparing actual vs desired outputs in applications like digit recognition.
Describes unsupervised learning by analogy to natural learning, clustering similar patterns.
Outlines unsupervised learning where no targets are provided, suitable for clustering tasks such as document grouping.
Explains self-organizing aspects in unsupervised learning where networks find patterns without error feedback.
Provides an overview of reinforcement learning where targets are absent, but guidance is provided through error feedback.
Explains the reinforcement learning process and how feedback signals guide learning based on evaluative criteria.
Lists various supervised and unsupervised learning algorithms to be covered, including Adaline, Perceptron, and Hebbian learning.
Introduces the perceptron learning rule and its relationship to error-signaling for adjusting weights.
Explains Hebbian learning as unsupervised learning focused on neuron connectivity changes based on patterns.
Describes how Hebbian learning alters connection strengths based on concurrent neuron activation.
Presents a practical example illustrating weight updates using the Hebbian learning rule.
Shows iteration examples of weight updates in Hebbian learning with specific calculations.
Application of Hebbian learning to implement an OR function by starting with zero weights.
Weight updates in successive iterations as per the Hebbian learning rule for reaching desired outputs.
Final adjustments in weight updates correlating actual and desired outputs within the Hebbian learning context.
More Related Content
Similar to Introduction to Neural networks (under graduate course) Lecture 7 of 9
Introduction to Neural networks (under graduate course) Lecture 7 of 9
- 1. Neural Networks Dr. RandaElanwar Lecture 7
- 2. Lecture Content • NNLearning techniques: – Supervised learning – Unsupervised learning – Reinforcement learning • Other learning laws: Hebbian learning rule 2Neural Networks Dr. Randa Elanwar
- 3. Basic models ofANN 3Neural Networks Dr. Randa Elanwar Basic Models of ANN Activation function Interconnections Learning rules
- 4. Learning • It’s aprocess by which a NN adapts itself to a stimulus by making proper parameter adjustments, resulting in the production of desired response • Two kinds of learning – Parameter learning:- connection weights are updated – Structure Learning:- change in network structure 4Neural Networks Dr. Randa Elanwar
- 5. Training • The processof modifying the weights in the connections between network layers with the objective of achieving the expected output is called training a network. • This is achieved through – Supervised learning – Unsupervised learning – Reinforcement learning 5Neural Networks Dr. Randa Elanwar
- 6. Supervised Learning • Childlearns from a teacher • Each input vector requires a corresponding target vector. • Training pair=[input vector, target vector] Neural Network W Error Signal Generator X (Input) Y (Actual output) (Desired Output) Error (D-Y) signals 6Neural Networks Dr. Randa Elanwar
- 7. Supervised learning 7Neural NetworksDr. Randa Elanwar Supervised learning does minimization of error
- 8. Supervised Learning • Learningis performed by presenting pattern with target • During learning, produced output is compared with the desired output – The difference between both output is used to modify learning weights according to the learning algorithm • Recognizing hand-written digits, pattern recognition and etc. • Neural Network models: perceptron, feed-forward, radial basis function, support vector machine. 8Neural Networks Dr. Randa Elanwar
- 9. Unsupervised Learning • Howa fish or tadpole learns • All similar input patterns are grouped together as clusters. • If a matching input pattern is not found a new cluster is formed 9Neural Networks Dr. Randa Elanwar
- 10. Unsupervised Learning • Targetsare not provided • Appropriate for clustering task – Find similar groups of documents in the web, content addressable memory, clustering. • Neural Network models: Kohonen, self organizing maps, Hopfield networks. 10Neural Networks Dr. Randa Elanwar
- 11. Self-organizing • In unsupervisedlearning there is no error feedback • Network must discover patterns, regularities, features for the input data over the output • While doing so the network might change in parameters • This process is called self-organizing 11Neural Networks Dr. Randa Elanwar
- 12. •Target is provided,but the desired output is absent. •The net is only provided with guidance to determine the produced output is correct or vise versa. •Weights are modified in the units that have errors Reinforcement Learning 12Neural Networks Dr. Randa Elanwar
- 13. Reinforcement Learning NN W Error Signal Generator X (Input) Y (Actual output) Error signalsR Reinforcement signal 13Neural Networks Dr. Randa Elanwar
- 14. When Reinforcement learningis used? • If less information is available about the target output values (critic information) • Learning based on this critic information is called reinforcement learning and the feedback sent is called reinforcement signal • Feedback in this case is only evaluative and not instructive 14Neural Networks Dr. Randa Elanwar
- 15. Some learning algorithmswe will learn are • Supervised: • Adaline, Madaline • Perceptron • Back Propagation • multilayer perceptrons • Radial Basis Function Networks • Unsupervised • Competitive Learning • Kohenen self organizing map • Learning vector quantization • Hebbian learning 15Neural Networks Dr. Randa Elanwar
- 16. Some learning algorithmswe will learn are • The only NN learning rule we have studied so far is the “perceptron learning” also known as “Error-Correction learning” or “delta rule”. • The learning signal is equal to the error signal: difference between the desired and the actual neuron output • We will consider also “Hebbian learning” and “Competitive learning” 16Neural Networks Dr. Randa Elanwar
- 17. Hebbian learning Rule •Feedforward unsupervised learning also known as “coincidence learning” • The learning signal is equal to the neuron’s output • To update the weights of a neuron i, the inputs to neuron i come from a preceding neuron j (xj) wi,j = wi,j + wi,j wi,j = * outputi * inputj wi,j = oixj • It is clear that Hebbian learning is not going to get our Perceptron to learn a set of training data, since weight changes depend only on the actual outputs and we don’t have desired outputs to compare to. • Hebb rule can be used for pattern association, pattern categorization, pattern classification and over a range of other areas 17Neural Networks Dr. Randa Elanwar
- 18. Hebbian learning Rule •If oixj is positive the results is increase in weight else vice versa • In other words: 1. If two neurons on either side of a connection are activated simultaneously (i.e. synchronously), then the strength of that connection (weight) is selectively increased. 2. If two neurons on either side of a connection are activated asynchronously, then that connection is selectively weakened or eliminated. • so that chance coincidences do not build up connection strengths. 18Neural Networks Dr. Randa Elanwar
- 19. Hebbian learning Rule •Example: Consider a 4-input perceptron that uses the bipolar binary sgn function +1 if net>0 sgn(net) = -1 if net<0 • to compute the output value o. If the weight vector w1 = (1 -1 0 0.5) and given that the learning constant = 1: • Use the Hebbian learning rule to train the perceptron using each of the following input vectors: • x1 = (1 -2 1.5 0)t • x2 = (1 -0.5 -2 -1.5)t • x3 = (0 1 -1 1.5)t 19Neural Networks Dr. Randa Elanwar
- 20. Hebbian learning Rule •Update weight vector for iteration 1 • Update weight vector for iteration 2 20Neural Networks Dr. Randa Elanwar 11,3 0 5.1 2 1 ]5.0011[1.)0( oXW 5.0 5.1 3 2 1.. 1)0()1( XOWW TT 12,75.0 5.1 2 5.0 1 ]5.05.132[2.)1( oXW 2 5.3 5.2 1 2.. 2)1()2( XOWW TT
- 21. Hebbian learning Rule •Update weight vector for iteration 3 21Neural Networks Dr. Randa Elanwar 13,3]5.1110.[ 5.1 1 1 0 ]25.35.21[3.)2( oXW 5.0 5.4 5.3 1 3.. 3)2()3( XOWW TT
- 22. Hebbian learning Rule •Example: OR function implementation • Use bipolar data in the place of binary data • Initially the weights and bias are set to zero w1=w2=b=0 22Neural Networks Dr. Randa Elanwar X1 X2 B Y (desired) 1 1 1 1 1 -1 1 1 -1 1 1 1 -1 -1 1 -1
- 23. Hebbian learning Rule •Update weight vector for iteration 1 • Update weight vector for iteration 2 23Neural Networks Dr. Randa Elanwar 11,0 1 1 1 ].000[1.)0( oXW 1 1 1 1.. 1)0()1( XOWW TT 12,1 1 1 1 ].111[2.)1( oXW 2 0 2 2.. 2)1()2( XOWW TT Ok Ok
- 24. Hebbian learning Rule •Update weight vector for iteration 3 • Update weight vector for iteration 4 • Since we are given desired outputs we have to correlate them to the actual outputs i.e. o4 = +1*ydes=-1 24Neural Networks Dr. Randa Elanwar 13,0 1 1 1 ].202[3.)2( oXW 3 1 1 3.. 3)2()3( XOWW TT Ok 14,1 1 1 1 ].311[4.)3( oXW 2 2 2 4.. 4)3()4( XOWW TT Wrong