Project ideas (see newest projects at the end – last updated 4/9/3 )
Each project needs
- Question: A project should have a clear, testable aim.
- Data: real-world data, or synthetic data generated by a program (either written by you or someone else)
- References: we’ve listed some references as starting points. You are expected to find others, especially if someone has worked on your question before.
- Tasks/analysis: Complex systems analyses involving structure, dynamics and function of your system
- Writeup: Clear description of the study, and a critical discussion of the implications of your results (see the project mark sheet 2b and 2c for headings for the final report).
Example projects. If you have another project in mind, discuss it with the course convenors.
Some of these projects are simple, others much more demanding. If you choose a simple project, you will be evaluated on how far you can push the analysis of the results with respect to issues in the literature and in what detail.
1. Network analysis (choose a real-world system and analyse the network structure)
a. Question: In your system, what are the nodes, what are the links between nodes? Are there alternative representations? How do the network properties differ with different representations?
b. Data:
Real world network, eg the world wide web, a social network, biological
interactions, ecosystems,
The data should be readily available (no credit will be given for collecting
the data in this course). It should be large enough to display interesting properties,
and available in a form that is analysable.
c.
References:
Strogatz, S.H. (2001). Exploring Complex
Networks, Nature 410, p. 268-276 (local
copy).
d.
Format
the data, represent it as a network using a program such as Pajek,
measure the statistics of the network. Compare the statistics for different
representations, and interpret your results.
2. Neural network models (hippocampus, visual cortex (eg V1, MT, MST), cerebellum, or other brain region for which a published model of the NN structure is available)
a. Question: How does the neural network structure of hippocampus affect its dynamics and functionality?
b. Data: O’Reilly’s model of CA1 and CA3
c. Reference(s):
d. Set up a random NN with the specified number of nodes and links. Analyse the connectivity, network statistics and dynamic behaviour of the untrained network. [advanced] train the network using a Hebbian learning rule and repeat the analysis.
A similar analysis for recurrent network models, such as J. Elman’s Simple recurrent network for language learning, or D. Floreano’s spiking neural networks used for robotic control.
3. Agent-based models (prisoner’s dilemma 1)
a. Question: What’s the difference between synchronous and asynchronous updating in a model of the prisoner’s dilemma?
b. Data/program: Set up an Axelrod-style competition between agents with different strategies.
c. Reference(s): Axelrod, R. (1984). “The evolution of cooperation”; Frean, M. (19xx) “” Journal of Theoretical Biology.
d. Track the proportion of different strategies in the population as the population evolves and compare the different patterns when updating is synchronous and when it is asynchronous.
Agent-based models (prisoner’s dilemma 2)
a. Question: How does a network of cooperation evolve in a population of prisoner’s dilemma agents [good programming skills required]
b. Data/program: Set up an Axelrod-style competition between agents with different strategies, with the difference that individual agent’s can choose whether to participate in any given interaction.
c. Reference(s): Axelrod, R. (1984). “The evolution of cooperation”; Batali, J.
d. Track the network of interactions at regular points in time during evolution.
Other prisoner dilemma studies are possible, see Matt Ridley’s “The Origins of Virtue”
4. Genetic regulatory networks
a. Question: What’s the effect of network structure on the dynamics of a random Boolean network?
b. Data/program: Set up an RBN with Kauffman’s random connectivity structure.
c. References: Kauffman, S. At Home in the Universe; find papers that reference RBNs and different connectivity structures.
d. Tasks: Compare the dynamics of Kauffman’s original network with an equivalent one with scale free connectivity.
Other issues could include adding stochastic or temporal components.
5.
Pattern
formation (pdf)
6.
L-systems
applications: See Jim Hanan to discuss these further jim@cpai.uq.edu.au (away 8-14 Sept)
http://www.cpai.uq.edu.au or
email Michael Renton mrenton@cpai.uq.edu.au
1. Simple models of environmental pressures on
plant structure.
2. Modelling insect damage as an emergent property of behaviour
3. Emergent flowering patterns as a result of plant signals
4. Patterns of spray deposits in a plant canopy.
http://www.cpai.uq.edu.au/images/our_research/virtual_plants/plant_animations/bugtime.mov
*** New (
7.
Artificial
Life based on Cellular Automata: models of physical, chemical, biological,
evolutionary and social systems.
Using L-studio, (or Repast, Swarm, or other
sufficiently powerful modelling system) replicate an Alife model to study an aspect of the real world.
References: , Artificial Life (journal), ALife and ECAL conferences; S. Wolfram, A new kind of science.
8.
…