To learn how to use the LMS
and Newton’s method to train a single-layer network
(2 points) Repeat the
computer experiment mentioned in the class by using Least-mean-square algorithm,
this time, however, positioning the two moons Figure to be on the edge of
separability, that is, d=0. Determine the classification error rate produced by
the algorithm over 2,000 test data points.
(3 points) Select one classification dataset from UCI first. Then design a single-layer network trained by Newton’s method. Provide your testing accuracy on the selected dataset.
The codes may upload to
Moss(https://theory.stanford.edu/~aiken/moss/) to check the similarity. If high
similarity rate is found, the assignment
will be marked as zero. The instructor will also report it to the Departmental Chair for a further penalty.
Get Free Quote!
330 Experts Online