Connectionism 2

Reprise: difficulties with representational theory of thought.

The key to learning in a connectionist system: backpropogation.

(Simulation of?) goal-seeking in a computer.

Hill-climbing algorithm

• Measure your distance from the summit.
• Take a step in an arbitrary direction.
• Measure your distance now.
• If you are nearer, take a step in the same direction as before.
• If you are not nearer take a step in another direction, chosen at random.

Representation in a connectionist system

'...the network had to create new, internal representations in the hidden layer of processing units. How did the network organise its "knowledge"?' Churchland and Sejnowski Reader p.143

Connectionism invites interdisciplinarity

The von Neuman model suggests: problem analysis, comutational algorithm, implementation. On this view, 'computational problems could be analysed independent of an understanding of the algorithm that executes the computation, and the algorithmic problem could be solved independent of an understanding of the physical implementation.' Churchland and Sejnowski Reader p.150.