Transcript Environmental challenges

NEW TIES year 2 review
NEW TIES =
New and Emergent World models
Through Individual, Evolutionary and Social learning
Timetable










10.00
10.20
10.45
11.15
11.30
12.00
12.30
13.00
14.00
14.20
14.50
–
–
–
–
–
–
–
–
–
–
–
10.20
10.45
11.15
11.30
12.00
12.30
13.00
14.00
14.20
14.50
15.30
Coordinator’s opening and summary
WP1 presentation: scenarios and challenges
WP2 presentation: evolving NEW TIES agents
Coffee break
WP3 presentation: language evolution and communication
WP4 presentation: data analysis tools
WP5 presentation: distributed NEW TIES platform
Lunch break
WP6 presentation: integration & evaluation
Questions and answers session
Review panel: deliberation (incl. coffee),
project participants: coffee break
15.30 – 16.00 Review panels feedback to project participants
What is NEW TIES?
An artificial agent world with
 Interesting scenarios / challenges
 Emergence engine =






Evolutionary learning
Individual learning
Social learning
Language evolution — link with IL & SL
Detection of world models (culture, data mining)
Large scale: many & complex agents, long simulations
Main objectives from Annex I
1.
2.
3.
To develop an artificial society with an emergent
culture.
To realise a powerful “emergence engine” as a
combination of individual learning, evolutionary
learning, and social learning.
To develop, evaluate, and use a range of social
learning mechanisms that allow sharing knowledge
with other members of the population.
Essential & distinguishing feature: enormous scale-up
NEW TIES questions (examples)






Can a NT society learn “ecologically correct”
behavior?
Can individual learning compensate for bad genes?
And social learning?
Can the agents develop language and share info
through it? Can we understand it?
Will telepathy work as social learning mechanism?
What culture will emerge?
Can we start a (p2p) SIG where users compete by
their “home-brewed tribes” in a NT world? Could we
win such a competition?
Modular Design
Agents
Language
Learning
Environment
New Ties
Virtual
Machine
Visualisers
Data Miners
Project structure (tech part)
Simulated world
WP5: p2p
infrastructure
WP1: environment &
challenges
WP3: language,
communication, cooperation
WP4: emerging
world models
WP2: agents and
learning
WP6: integration
and evaluation
Year 2 in brief


Major code restructuring  effective start of NEW
TIES experiments in April-May 2006
Experiments:

Evolutionary learning:
 simple world (calibration) and
 poison world (challenge solved)


Language evolution: collective lexicon developed
Development:



Scenario generator
World model detectors, data analysis (user in the loop!)
Distributed platform
Main achievements per WP







WP1:
WP2:
WP3:
WP4:
WP5:
Scenario generator and map viewer
Evolutionary learning in NEW TIES
Language evolution in NEW TIES
Interactive data analysis tools
Distributed platform beta, incl.
historical data module
WP6: Complete code restructuring
WP7: Release of the NEW TIES platform
Biggest challenges at the moment




Evolution remains the only learning mechanism, i.e.,
no IL and no SL
Evolved language (components) not used by agents for
info exchange or as building blocks in IL
Simulation times are too long
No challenging and appealing scenario solved
NEW TIES must become more than another ALife project