Knowledge Test Answers

1.   

• The Neuron's output is either on or off.

• The output depends only on the inputs. A certain number must be on at any one time in order to make the neuron fire.

• Each input has a multiplicative weight on it into the neuron. A bigger weight represents a strong signal being transmitted, and a lower weight represents a weak signal being transmitted.

2.    Feedforward.
3.    Activation.
4.    Unit Step Function.
5.   

6.    Supervised Learning.
7.    Allows the neuron to learn from its mistakes.
8.    step 1) Intiliase threshold and weights.
       step 2) Present input and desired output.
       step 3) Calculate the actual output.
       step 4) Adapts weights.
9.    Positive gain function between zero and one.
10.  It ends when a user specified error threshold becomes greater
       than the iteration error OR when a predefined number of iterations are reached.
11.  A single layer perceptron can solve "linearly seperable" problems as shown below

12. 

Sigmoid: f(x) = (1 + e^-ßx)^-1

13. 

14. "the backpropagation rule" OR "the generalised delta rule".
15.

When showing the untrained network an input pattern, it produces a random output. We need to define a error function which represens the different between our output recieved and the output we desire. This error function must be continually reduced so our output reaches that of the desired output.

To achieve this we adjudt the weights on the links between layers, this is done by the generalised delta rule calculating the error for a particular input and back-propagating it through to the previous layer. Thus each unit in the network has its weights adjusted so that it reduces the value of the error function and the network learns.

16. A multi layer perceptron may solve either linearly seperable and inseperable problems of the kind below, which is unsolveable by the
      SLP:

Previous:- Knowledge Test Questions

Next:- Home