Book contents
- Frontmatter
- Contents
- 1 Introduction
- 2 Self-organisation in complex systems
- 3 Network evolution and the emergence of structure
- 4 Artificial life: growing complex systems
- 5 Deterministic and random fractals
- 6 Non-linear dynamics
- 7 Non-linear control systems
- 8 Parallel computers and complex systems
- 9 Are ecosystems complex systems?
- 10 Complexity and neural networks
- Index
10 - Complexity and neural networks
Published online by Cambridge University Press: 04 August 2010
- Frontmatter
- Contents
- 1 Introduction
- 2 Self-organisation in complex systems
- 3 Network evolution and the emergence of structure
- 4 Artificial life: growing complex systems
- 5 Deterministic and random fractals
- 6 Non-linear dynamics
- 7 Non-linear control systems
- 8 Parallel computers and complex systems
- 9 Are ecosystems complex systems?
- 10 Complexity and neural networks
- Index
Summary
Introduction
Neural networks fuse together ideas from all aspects of complex systems: they embody features of all the other chapters of the book; even the ecology of neural networks, the relationship of brain to animal environmental niche, is a rich and exciting field (Snyder et al., 1990). In themselves neural dynamics are rich in complexity. But animal neural networks, brains, play a fundamental role in the characterisation, prediction and control of many phenomena of the natural and man-made world. They evolved as the monitors, memory and controllers of natural complex systems. We should know how they work.
Three features of biological brains create diverse and fascinating behaviour:
many autonomous agents: the human brain has 109 neurons, about the number of people on earth and considerably larger than, say, the largest ant colonies. Each agent (neuron) is adaptable.
vast interconnectivity: the human brain contains of the order of 1014 synapses, the points of connection between neurons. Each neuron may connect with as many as 105 other neurons.
each agent (nerve cell or neuron) learns. Biological brains are the most complex adaptive system of which we know. Their adaptability not only serves to cope with a changing environment but also to provide substantial redundancy to damage.
These factors alone would be enough to generate vast complexity, yet each neuron is, itself, a highly sophisticated biochemical system, made up of many interacting components. In many complex systems, we can encapsulate the interior workings of each agent, i.e. handle only external behaviour and interactions. Unfortunately, as we shall discover in this chapter, it is not obvious that we can do this with biological neural networks.
- Type
- Chapter
- Information
- Complex Systems , pp. 367 - 406Publisher: Cambridge University PressPrint publication year: 2000
- 1
- Cited by