Artificial Neural Network

Artificial Neural Network

1. Massive distributed parallel processor

   • Capable of emulating human reasoning
   • They are not algorithmical

2. Components

   • Input
     • Sensors
     • Information
   • Associative hidden layers
     • Single layer
     • Multilayer
     • Recurrent networks
   • Output

3. Brain vs machine

   • Speed
     • Silicon logic gates are 5 times quickier
     • As a whole the brain is faster because the quantity of connections
   • Energy
     • Computer: 10-6 joules
     • Brain 10-16 joules

4. Learning paradigms

   • Supervised learning
   • Reinforcement learning
   • Unsupervised learning