knowledge pt 4 Flashcards
Connectionism:
what is Parallel distributed processing (PDP)
what are its Key components?
computer models for representing cognitive processes
loosely based on neurons and neural networks
UNIT: in layer (neurons)
CONNECTIONS: between units (synapses)
Connectionist network
outer layer
receive input from hidden layers
provide resulting signals from the network
Connectionist network
Hidden layer
recieve input from input and hidden layer
send output to hidden layers and output layer
Connectionist network
input layer
activated by stimulation from environment
send signal to hidden layer
Connectionist network
each conection has a connection weight. what is that?
strength of connection between two units
determines how strongly an incoming signal will activate the next unit
changes in connection weights typically happen slowly, and represent learning and memory
like the strength of synaptic connection
Connectionist network
each unit has an activation
what does it represent and changes do what?
represents how active the neuron currently is
changes in activation happen quickly, and represent moment-tomoment information processing
firing rate of neuron
what is supervised learning?
how learning occurs in a connectionist network
Supervised learning:
what is error signal?
difference between actual activity of each output unit and correct activity
supervised learning
what is backpropogation?
algorithm to update connection weightts based on error signal
- updating starts at the output units and works backwards towards the input unit
In supervised learning of concepts with McMlelland and Roger
after backprop what with learning ?
properties are correctly associated with concepts
In supervised learning of concepts with McMlelland and Roger
after backprop what with similarity?
similar concepts share similar representations
In supervised learning of concepts with McMlelland and Roger
after backprop what with Generalization
knowledge is shared amount similar concepts
In supervised learning of concepts with McMlelland and Roger
after backprop what with Graceful degradation
performance decreases gradually as more and more of the network is damaged
Categories in the brain
sensory functional hypothesis?
organization based on sensory attributes
Categories in the brain
semantic category/domain-specific approach
organization based on shared features
