Unformatted text preview: n tree algorithm on the same
graph. Describe in one sentence (or function) how the decision boundaries translate to walls on the original, xy-plane graph. 6 Question 2: Neural Networks and Genetic
Algorithms (50 points)
Part A (20 points)
Perceptrons are the basic units of neural networks as we have seen them. They take a list of inputs x,
multiply them by a list of corresponding weights w, compute the sum of those products, and pass the
result through a decision function. We use a "fake" input T, usually -1, times an associated weight wT,
as part of the sum. When using a single perceptron for classification, one usually uses a threshold
decision function: if the sum z > 0, the output is 1, otherwise 0. x1
w3 -1 wT ∑ wn To explore what perceptrons can and cannot do, we will ask you to make up weights, rather than
training them. Consider the boolean function A→B, and note that it is the same as ¬A∨B.
1. Can a single perceptron with inputs A and B output 1 iff A→B?
If yes give weights: wA=_______ wB=______...
View Full Document
- Fall '10
- Artificial Intelligence, Artificial neural network, neural network, Shaun, decision function