Artificial Neural Networks
Artificial Neural Networks
Artificial Neural Networks
Introduction
OUTLINE:
• - Artificial Neural Networks: Properties, applications
- Artificial Neuron
- Perceptron
• Adaptivity
–changing the connection strengths to learn things
• Non-linearity
–the non-linear activation functions are essential
• Fault tolerance
–if one of the neurons or connections is damaged, the
whole network still works quite well
Properties of ANN:
They might be better alternatives than classical
solutions for problems characterised by:
–Nonlinearities
–High dimensionality
–Noisy, complex, imprecise, imperfect and/or
error prone sensor data
–A lack of a clearly stated mathematical solution
or algorithm
Artificial Neural Network: