2. Artificial Neural Network (ANN)
• A mathematical model inspired by biological
neural networks
• ANN consists of an interconnected group of
artificial neurons
▶
3. Compare to Brain
Biological Neural Network Artificial Neural Network
Neuron Unit (or node)
Synapse Connection
Inhibition or Excitation of Neuron Connection Weight
Threshold of firing rate Activation Function
4. ANN Structure
ANN BNN
Input Layer Sensory Neurons
Hidden Layer Interneurons
Output Layer Motoneurons
5. Attractions of ANN Model
• Learning
– Human brain can learn by changing their
interconnections between neurons
– ANN can learn by changing their connection
weights between units
• Parallel Processing : Many processes simultaneously
• Robustness: It works even if it is damaged
7. Example – AND Operator
Unit
1
Unit
3
Unit
2
X f(X) Y
0 0 0 0 0
0 1 0.5 0.5 0
1 0 0.5 0.5 0
1 1 1 1 1
8. Example – OR Operator
Unit
1
Unit
3
Unit
2
X f(X) Y
0 0 0 0 0
0 1 0.5 0.5 1
1 0 0.5 0.5 1
1 1 1 1 1
9. Example – XOR Operator
Unit Unit
1 3
Unit
5
Unit Unit
2 4
Unit 1 Unit 2 Unit 5
X f(X) Y
0 0 0 0 0
0 1 0.5 0.5 1
1 0 0.5 0.5 1
1 1 0 0 0
10. How to Learn?
1. Set all connection weight as randomly
2. Input the data
3. If the output corrects (expected value)
▷ Then, exit iteration
▷ Else, change the connection weights to reduce
difference and repeat (go to 2)
How to change connection weight?
There are many algorithms but is hard to explain because of the
margin of the slide is too small!!
11. Applications
• Pattern Recognition
– Voice Recognition
– Medical Treatment (e.g. cancer detect)
• Data Processing
– Noise Filtering
• Robotics
– Data-Driven Predictive Controller