代做ECEN 649: Patten Recognitions 、Matlab编程代做留学生、代做PYTHON语言、R编程代写留学生-代写Web作业

ECEN 649: Patten Recognitions

Assignment 1 – Fall 2018

Due date: Thursday September 27, 2018

Reading assignment: Lecture notes 1 & 2 (required), Chapter 1, 2, and 9 of the textbook

(recommended), Wikipedia page on Perceptron (recommended)

Computer assignment: For this assignment, please do NOT use any existing software

packages for naive Bayes or Perceptron.

(Naive Bayes) Download the MNIST data sets in the CSV format at https://pjreddie.

com/projects/mnist-in-csv/. Each sample in the data sets is a 28 × 28 binary image of

a hand-written digit (0–9). One sample from the data set is depicted in Figure 1.

Figure 1: A hand-written digit.

1) Use the train set to estimate the probabilities for getting each digit and the conditional

probabilities of individual pixels given the values of the digit. Record your results into

a table.

2) For each sample in the test set, use the probabilities estimated from the first part and

Naive Bayes to predict the value of the digit based on the values of all 28 × 28 = 784

pixels. Calculate the success rate of your predictor.

(Perceptron) Download the IRIS data sets in the CSV format from your email. Each sample

from the data sets includes two real-valued features and one binary outcome.

1) Plot each sample from the train set in a 2-D plot. Use the Perceptron algorithm to

learn a half space predictor that completely separates the positive samples from the

negative ones. Plot the decision boundary of your predictor on the same 2-D plot.

2) Plot each sample from the test set and the decision boundary of your predictor in a

separate 2-D plot. Calculate the success rate of your predictor.