Transcript Slide 1

MODELING AND VISUALIZING
DYNAMIC ASSOCIATIVE
NETWORKS:
Towards Developing a
More Robust and
Biologically-Plausible
Cognitive Model
Based on Dr. Anthony Beavers’
ongoing research
By Michael Zlatkovsky, dual-major in
Computer Science and Cognitive Science
I’M A PC...
I’m a neural net
WHY NEURAL NETS?
Pattern recognition
 Inferring a function by observation
 Robustness against errors
 Parallel nature

ARTIFICIAL NEURAL NETWORKS
ARTIFICIAL NEURAL NETWORKS

Artificial way of adjusting: setting weights
DR. BEAVER’S DYNAMIC ASSOCIATIVE
NETWORK MODEL

Dr. Beavers,
Director of UE’s
Cognitive
Science
Department, is
attempting to
explore a
different model
of cognition.
DR. BEAVER’S DYNAMIC ASSOCIATIVE
NETWORK MODEL
No more mystery
“hidden layer”
 Learning through the
order and structure of
experience



No “unnatural”
training
Organic network

Can incorporate new
information
DAN’S COGNITIVE ABILITIES COME FROM
LONG-TERM LEARNING AND CURRENT STATE
TRANSLATION INTO A NODE-CENTRIC
MODEL
EARLY EXCEL PROTOTYPE
THE DAN SOFTWARE SUITE

Based on prototype, create a self-contained DAN
Model

Written in Java;
object-oriented approach



Expand on features of Excel Model
(various activation modes, learning
mode, settings)
Most importantly: focus on
design fundamentals to ensure
speedy operation and high
capacity.
Create visualization routines
RE-CALCULATIONS

Most frequent operations

DANs are massively parallel
Re-computing from scratch: O(n2).
 EX: for 1000 node-network, change
in 2 nodes that impact 5 nodes each...



Instead of 10 re-calculations, 1,000,000!
My scheme: buffered change-propagating
dependency-driven re-calculations
OTHER DESIGN CONSIDERATIONS



General separation of concerns (59 classes)
Model-View-Controller
 “Core framework” with “helper” controllers &
GUI views/wrappers
GUI look, cross-platform
VISUALIZATION
PREFUSE
framework
 Radial tree layout
(PREFUSE)
 Color nodes based on
activation
 Color edges based on
connection type
 Highlighting,
animation, etc.

RESULTS: DAN SOFTWARE SUITE

Overall successful
 Quick
 Convenient UI
 Adaptable
 True to model
RESULTS: DAN MODEL

Promising results: various rudimentary
cognitive abilities:
 “Initial Intelligence”: pattern recognition,
feature detection, memorization of simple
sequences, identification of similarities and
differences, storage of relational data,
comparison and classification, etc.
 Possibly, building blocks of more sophisticated
intelligence.
RESULTS: DAN MODEL

Has not gone unchanged:
RESULTS: DAN MODEL

Has not gone unchanged:
RESULTS: DAN MODEL

Has not gone unchanged:
training:
“the boy woke up”
“the boy fell asleep”
“the boy woke up”
“the boy fell asleep”
RESULTS: DAN MODEL

Has not gone unchanged:
training:
“the boy woke up”
“the boy fell asleep”
“the boy woke up”
“the boy fell asleep”
RESULTS: OVERALL


More robust?
 Don’t know... Yet.
 Received with curiosity and some enthusiasm
by researchers working in the field.
More biologically plausible?
 Absolutely.
 Hebbian Neurological Principle: nodes that
“fire together, wire together”.
 Contrast with ANNs’s statistically-based
learning
I’m a DAN