perceptrons
the very beginning of artificial neural networks
How does real neurons propagate a signal ?
a really simplified explanation
Donald Hebb
(1904 -1985)
Hebbian theory
“When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased”
The Organization of Behavior
Hebb, D.O. (1949)
"Cells that fire together, wire together."
Siegrid Löwel
initial state
simultaneous firing
initial state
final state
“(...) adults who have never played the piano fail to activate their pre-motor cortex to the sound of piano; after five one-hour piano lessons, however, they do, suggesting that the repeated association between the sound and the action of pressing a key has created a Hebbian association between auditory neurons and premotor MNs”
Programmed to learn? The ontogeny of mirror neurons
Marco Del Giudice, Valeria Manera and Christian Keysers
Developmental Science - 2009
Mad scientist from 1960
i1
i2
i3
in
output
w1
w2
w3
wn
transfer function
i1
i2
i3
in
w1
w2
w3
wn
output
features
outputs
input
layer
hidden
layer
output
layer
supervised learning:
back-propagation algorithm
example !
excelent
good
fair
cattle blastocysts
poor
image segmentation
Total of 24 features, after dimensionality reduction
training: 70%
validation: 15%
test: 15%
(only 98 samples !)
multi layer perceptron
deep learning
