AIMS
OF TODAY’S SESSION
·
to
go over the main points of the lecture and workshop
·
reminder
on Lab Report
·
familiarization
with Connectionist/PDP models
·
evaluate
relative merits of Symbolic and Connectionist approaches
·
next
week: ACT* and Soar
QUIZ on Neural Networks
(1)
In connectionist
or PDP models, what is a ‘unit’? What does it correspond to in real people?
(2)
Name three kinds
of units. Roughly what does each
kind do?
(3)
What is a
‘layer’?
(4)
In the model that
you have studied, are there connections between layers, within layers or both?
(5)
What’s the
difference between a feed-forward and a recurrent network? (look at Green book)
(6)
What is a
connection weight?
(7)
What are the two
kinds of connection?
(8)
What does PDP stand
for? What is the significance of
the ‘D’?
(9)
What is the
simple rule for firing of a unit which was discussed in lecture?
Input
units 1 – 4 are are all connected to hidden units H1 and H2.
Hidden
unit H1: threshold 0.2
Input
units:
(1)
Activation: 0, weight
0.3
(2)
Activation: 1,
weight 0.7
(3)
Activation: 0,
weight 0.9
(4)
Activation: 1,
weight 0.4
Hidden
unit H2: threshold 0.6
Input
units:
(5)
Activation: 1,
weight -0.3
(6)
Activation: 0,
weight 0.7
(7)
Activation: 1,
weight 0.9
(8)
Activation: 0,
weight -0.4
Do
either H1 or H2 fire?