Assignment
1. Consider a binary classifier which assigns pattern X to class โOโ
(positive class) if f (x) > 0 and to class โXโ (negative class) if
f (X) < 0, where
f (x) = a + bx + cx2 (37)
Observe, based on the discussion above, that equivalently we can
assign X to โOโ if zt Xโ > 0 and to class โXโ if zt Xโ < 0, where
Let us consider a set of labeled patterns that are not linearly sep-
arable. Specifically, let us consider the one-dimensional data set
shown in the following table. Normalize and transform the data
and learn the weight vector using perceptron learning algorithm.
2. Consider a two-class two-dimensional dataset given by
Class1 : (1, 2)t, (1, 1)t, (2, 2)t and Class2 : (2, 0)t, (1,โ2)t.
(a) Obtain the support vectors.
(b) What is the Decision Boundary?
Docsity.com
