Define a function which implements equation (4). This function will only
receive the vector ⃗α as a parameter. You can use global variables for other
things that the function needs (t and K values).

Implement the indicator function (equation 6) which uses the non-zero
αi
’s together with their ⃗xi
’s and ti
’s to classify new points

In order to see your data, you can use the plot functions from matplotlib

Make the call to minimize as indicated in the code sample above.

This function should implement the equality constraint of (10). Also here,
you can make use of numpy.dot to be efficient.

If the data is well separated, only a few of the α values will be non-zero.
Since we are dealing with floating point values, however, those that are
supposed to be zero will in reality only be approximately zero. Therefore,
use a low threshold (10−5
should work fine) to determine which are to be
regarded as non-zero.
You need to save the non-zero αi
’s along with the corresponding data
points (⃗xi) and target values (ti) in a separate data structure, for instance
a list.

Note that you must use a point on the margin. This corresponds to a
point with an α-value larger than zero, but less than C (if slack is used).

Implement the two non-linear kernels. You should be able to classify
very hard data sets with these.