Plot the logistic loss or negative log likelihood

Assignment Help Computer Engineering

Reference no: EM132158983

USE R LANGUAGE AND MNIST.CSV In this question, you will experiment with different training set sizes and see how the accuracy changes with them. This question should be done for both 0,1 and 3,5 classification.

a. For each set of classes (0,1 and 3,5), choose the following sizes to train on: 5%, 10%, 15% ... 100% (i.e. 20 training set sizes). For each training set size, sample that many inputs from the respective complete training set (i.e. train_0_1 or train_3_5).

Train your model on each subset selected, test it on the corresponding test set (i.e. test_0_1 or test_3_5), and graph the training and test set accuracy over each split (you should end up with TWO graphs - one showing training & test accuracy for 0,1 and another for 3,5 set).

Remember to average the accuracy over 10 different divisions of the data each of the above sizes so the graphs will be less noisy. Comment on the trends of accuracy values you observe for each set.

b. Repeat 5a, but instead of plotting accuracies, plot the logistic loss/negative log likelihood when training and testing, for each size. Comment on the trends of loss values you observe for each set.

Reference no: EM132158983

Questions Cloud

Obtaining a white male : What is the probability, as a proportion, of obtaining a white male? [answers to 2 decimal places]

What is the conditional probability : A diagnosis for a test is such that it (correctly) detects the disease in 86% of the individuals who actually have the disease. Also, if a person does not have

Write a program that reads in a list of accounts from a file : Write a program that reads in a list of accounts from a file, and then allows the user to enter transactions for each account.

What is the decay rate : The average lifetime of a certain new cell phone is 3 years. The manufacturer will replace any cell phone failing within 2 years of the date of purchase.

Plot the logistic loss or negative log likelihood : Train your model on each subset selected, test it on the corresponding test set (i.e. test_0_1 or test_3_5)

Represent the sample mean level of support : The study provided the following information on the distribution of levels of support for? sustainability: µ=64?, s=23. Now consider a random sample

Displays the number of the files in the directory : Use a recursive approach to write a program that prompts the user to enter a directory and Displays the number of the files in the directory.

Compute greatest common divisor using recursion : Write a test program that prompts the user to enter an index and displays its Fibonacci number.

Write a program to convert time in days and hours : He wants you to write a piece of code that prompts the user for number of days, hours, minutes, and seconds (in that order).

User Account

All Pages