Task 3 Brains and Computers Flashcards
Connectionism
An artifical intelligence appraoch which models cognition in artifical neural networks.
Local representation
Nodes in a neural network each represent one representational element.
Distributed representation
Nodes in a neural network each represent seperate elements of a representational element.
Graceful degradation
Network breaks down only slowly as damage increases
- small damage has no effect on network performance
Activation functions
The function of a node in a neural network that determines the activity of a neuron. (Linear, Threshold linear, Binary, Sigmoid)
Output functions
The function of a node in a neural network that determines the output a node sends forward
Transfer functions
The function of a node in a neural network that determines the transformation on the input to create an output
Multilayerd network
A neural network that contains one hidden layer, and are useful for higher level computations
Deep neural network
A neural network that contains multiple hidden layers, and are useful for higher level computations
Feedforward network
A neural network in which information flows in one direction throught the network
Recurrent network
A neural network which exhibits temporal dynamic behaviour by sending its output back into itself as well as to the next layer
Backpropagation
- The delta rule applied to multi-layer networks propagates the error back through the neural network
- To find total error of neuron in hidden layer
Delta rule
- Method used to calculate the difference between actual and desired output (error) and change the weights accordingly (That the error equals 0)
- Gradual changes with each learning trial
Gradient descend learning
An optimization algorithm used in machine learning to find the optimal weight levels of nodes in a neural network. (Only find local optimum)
Neural network
An artificial neural network, composed of artificial neurons or nodes.
