ECS>Users>Marcus>

Marcus is in the Artificial Intelligence group and has interests in machine learning, complex adaptive systems, and evolutionary dynamics.

What's on this page:

Research
Teaching
Miscellaneous other stuff

Research

I'm interested in machine learning, and evolution.

Publications

all, 2011, 2010, 2009, 2008, 2007, 2006, 2005, 2004, 2003, 2002, 2001, 2000, 1999, 1998, before 1998

Machine Learning

Gaussian processes

	Gaussian process models in machine learning, specifically how they might be used to carry out efficient optimization in domains where data is expensive or risky to obtain. Gaussian process inference can be extended to model multiple, dependent, outputs - one nice application of this is that it allows "cheap" data to be used as a proxy for "expensive" data. Much of this work can be found in Phillip Boyle's PhD thesis.

Gaussian process models in machine learning, specifically how they might be used to carry out efficient optimization in domains where data is expensive or risky to obtain. Gaussian process inference can be extended to model multiple, dependent, outputs - one nice application of this is that it allows "cheap" data to be used as a proxy for "expensive" data. Much of this work can be found in Phillip Boyle's PhD thesis.

Graphical models

	I'm interested in inference in graphical models. Russell Tod looked at how spiking neurons might implement probability propagation in his MSc. I gave a tutorial on restricted Boltzmann machines and deep belief nets associated with the AI'09 artificial intelligence conference in Melbourne, and a (3 hour) tutorial at ANU's Australian Summer School for machine learning in 2010. video lectures.

I'm interested in inference in graphical models. Russell Tod looked at how spiking neurons might implement probability propagation in his MSc. I gave a tutorial on restricted Boltzmann machines and deep belief nets associated with the AI'09 artificial intelligence conference in Melbourne, and a (3 hour) tutorial at ANU's Australian Summer School for machine learning in 2010. video lectures.

Reinforcement learning

	Tim Field (now at Willow Garage) used the policy gradient algorithm to train local policies for motor control; James Bebbington is currently looking at reinforcement learning of policies for active sensing by deep belief nets

Tim Field (now at Willow Garage) used the policy gradient algorithm to train local policies for motor control; James Bebbington is currently looking at reinforcement learning of policies for active sensing by deep belief nets

Evolution

How network structure affects the rate of evolution

	I've been studying the effect that network structure has on the rate of evolution. This is work with Gareth Baxter (former postdoc at ECS, now in Portugal), Stephen Hartley (VUW's SBS) and Paul Rainey (Massey Albany)

I've been studying the effect that network structure has on the rate of evolution. This is work with Gareth Baxter (former postdoc at ECS, now in Portugal), Stephen Hartley (VUW's SBS) and Paul Rainey (Massey Albany)

Evolution of cooperation

	With Joseph Bulbulia I've been thinking about how humans are so good at forming ad hoc cooperative groups, and how this relates to possible evolutionary drivers behind our strong religious tendencies. I've had a long-standing interest in the Prisoner's Dilemma, and the evolution of enslavement - Edward Abraham and I studied how natural selection can warp mutually productive relationships into exploitative ones.
	Fight club... simulating the evolution of "ganging up"

With Joseph Bulbulia I've been thinking about how humans are so good at forming ad hoc cooperative groups, and how this relates to possible evolutionary drivers behind our strong religious tendencies. I've had a long-standing interest in the Prisoner's Dilemma, and the evolution of enslavement - Edward Abraham and I studied how natural selection can warp mutually productive relationships into exploitative ones.

Fight club... simulating the evolution of "ganging up"

Cyclic competitions in nature

	With Edward Abraham I have studied rock-scissors-paper situations in ecological systems. Such systems show paradoxical behaviour - for example the slowest invader is the one most likely to survive while its competitors go extinct - a phenomenon we dubbed "survival of the weakest". Richard Mansfield extended these ideas significantly in his PhD thesis. Richard and I showed how the 3-way competition could arise from even simpler systems of just two species, or even one.
	Losing to Peter Whigham for the papers

With Edward Abraham I have studied rock-scissors-paper situations in ecological systems. Such systems show paradoxical behaviour - for example the slowest invader is the one most likely to survive while its competitors go extinct - a phenomenon we dubbed "survival of the weakest". Richard Mansfield extended these ideas significantly in his PhD thesis. Richard and I showed how the 3-way competition could arise from even simpler systems of just two species, or even one.

Losing to Peter Whigham for the papers

Teaching

Grad students

Marcus graduate students
Possible thesis topics
the Festival of Doubt is an interdisciplinary discussion series on complex adaptive systems.

Courses in 2011

Machine learning (COMP421)
Introduction to Data Structures and Algorithms (COMP103)
COMP489/COMP588 projects: I am the course coordinator for these projects. Students can look at possible 30-point projects at https://ecs.victoria.ac.nz/cgi-bin/pas.py/home