Two Layer Artificial Neural Networks:

However decision trees are whenever powerful they are as a simple representation scheme. Whereas graphical on the surface that they can be seen as disjunctions of conjunctions then are a logical representation so we call such schemes symbolic representations. Generally in this lecture here we look at non-symbolic representation scheme that's known as ‘Artificial Neural Networks'. However this type term is often shortened to ‘Neural Networks' so hence this annoys neuro-biologists whom deal with such real neural networks as inside our human heads.

However remember there that ANNs have a biological motivation then we briefly look at that first. If there, we look in detail at how information is represented in ANNs and we look at the simplest type of network so then two layer networks are there. Surely we look at perceptrons and linear units then discuss the limitations that is so simple networks have. Thus in the next lecture here we discuss the back-propagation algorithm and multi-layer network for learning such networks.