Connectionism

• I work with models that attempt to be quite like neurons.
• However, there is some important differences between traditional computers (say a PC), and the brain.
• 500 billion processors vs. 1, but also a computer processor is much more complex, and fast than a neuron.
• The distributed computation of the brain is really different from the way we write programs.
• Parallel processing is a sub-field of computer science that attempts to solve the problems of writing parallel programs.
• Why do we need parallel processing?
• Typically, parallel processing deals with lots of processors doing the same thing.
• However, there is a field closer to neurons called connectionism where different parts do different calculations.
• These algorithms used to be called neural nets because they were inspired by the brain.
• The term connectionism is used now to refer to non-neural, highly distributed models.
• Connectionist systems are very simplified models of neurons and synapses.
• This means that we can understand them mathematically, and thus can get them to solve real world problems.