https://github.com/devstein/neural-net-practice
Creating a neural net from scratch for hw
https://github.com/devstein/neural-net-practice
Last synced: 4 months ago
JSON representation
Creating a neural net from scratch for hw
- Host: GitHub
- URL: https://github.com/devstein/neural-net-practice
- Owner: devstein
- Created: 2016-11-17T03:40:18.000Z (over 9 years ago)
- Default Branch: master
- Last Pushed: 2016-11-17T18:19:54.000Z (over 9 years ago)
- Last Synced: 2025-10-13T19:19:16.883Z (4 months ago)
- Language: Python
- Homepage:
- Size: 3.97 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README
Awesome Lists containing this project
README
#######################################################################################
# README FILE
# Introduction to Machine Learning
# Assignment 4
#
# After completing the assignment, fill in this README file. It asks you to duplicate
# your answer to several questions. Be certain that your answers in this README file
# (which will be used for auto-grading) match the answers in your PDF writeup.
#
# This README file is formatted in YAML to allow it to be machine readable. Please
# be very careful with how you edit it, being careful to following the formatting.
# Make certain not to change any text in all UPPERCASE. After editing, you can make
# certain that your README file follows proper YAML syntax by running it through an
# online YAML checker, such as http://yaml-online-parser.appspot.com/
#
# The most frequent error is multi-line strings, such as "SOURCESCONSULTED" and
# "FEEDBACK_ERRORS". Make certain that the previous line ends with a | followed by a
# newline. Then, make certain that the subsequent lines are indented one level in
# (4 spaces). This sounds complex, but just follow the existing format of this file.
# If you run into any problems that you can't fix easily in listing your sources or
# providing feedback on the assignment, just include your multi-line string answers to
# these parts as comments. Everything else (all simple one-word or numeric answers) must
# be properly formatted YAML.
#
# Assignment Version 20151026a
#######################################################################################
# Personal information (BE SURE TO CHANGE THIS!)
FIRSTNAME: Devin
LASTNAME: Stein
PENNKEY: devstein
PENNID: 17504818
# Which course are you enrolled in? (enter 419 or 519)
COURSE: 419
# List all sources of help that you consulted while completing this assignment
# (other students, colleagues, textbooks, websites, etc.). This includes anyone you
# briefly discussed the homework with. If you received help from the following sources,
# you do not need to cite it: course instructor, course teaching assistants, course
# lecture notes, course textbooks or other readings.
#
# If you didn't receive help from anyone, write "none".
SOURCESCONSULTED: |
While completing the assignment, I consulted the following sources:
Chapter 2 of Pattern Recognition and Machine Learning by C. Bishop
Chapter 5 of Machine Learning, A Probabilistic Perspective by K. Murphy
Coursera
#######################################################################################
# Answers to Problem 2: Backpropagation with Momentum
#
# Please list your final answers below for auto-grading.
#######################################################################################
# Problem 2: What is the weight vector after the first epoch? (only your final answer)
PROB_BACKPROPMOMENTUM_EPOCH1: (0.1, 0.1, 0.1, 0.1, 0.1)
# Problem 2: What is the weight vector after the second epoch? (only your final answer)
PROB_BACKPROPMOMENTUM_EPOCH2: (0.1, 0.1, 0.1, 0.1, 0.1)
#######################################################################################
# Answer to Implementation Exercise 1: Text Classification and ROC
#######################################################################################
# What was the training accuracy?
TEXT20NEWS_ACCURACY_TRAIN_NB: 0.959165635496
TEXT20NEWS_ACCURACY_TRAIN_SVM: 0.996199398975
# What was the testing accuracy?
TEXT20NEWS_ACCURACY_TEST_NB: 0.819437068508
TEXT20NEWS_ACCURACY_TEST_SVM: 0.846255974509
# What was the training precision?
TEXT20NEWS_PRECISION_TRAIN_NB: 0.963001641562
TEXT20NEWS_PRECISION_TRAIN_SVM: 0.996194255998
# What was the testing precision?
TEXT20NEWS_PRECISION_TEST_NB: 0.841965329356
TEXT20NEWS_PRECISION_TEST_SVM: 0.848552735505
# What was the training recall?
TEXT20NEWS_RECALL_TRAIN_NB: 0.94971825651
TEXT20NEWS_RECALL_TRAIN_SVM: 0.996127608867
# What was the testing recall?
TEXT20NEWS_RECALL_TEST_NB: 0.803871775123
TEXT20NEWS_RECALL_TEST_SVM: 0.839373845703
# What was the training time (in seconds)?
TEXT20NEWS_TIME_TRAIN_NB: 0.215321
TEXT20NEWS_TIME_TRAIN_SVM: 26.950086
# Which classifier performed better overall? (answer either NAIVEBAYES or SVM -- you can choose only one)
TEXT20NEWS_BESTCLASSIFIER: SVM
#######################################################################################
# Answer to Implementation Exercise 2: Neural Network
#######################################################################################
# What was the training accuracy?
NEURALNET_ACCURACY_TRAIN: 0.9552
# What was the optimal learning rate?
NEURALNET_OPTIMAL_LEARNINGRATE: 3.00
#######################################################################################
# Feedback on the Assignment (OPTIONAL)
#
# The following information will help us improve future versions of this assignment.
# It is completely optional, but highly appreciated. Please be honest.
#######################################################################################
# Approximately how many hours did it take you to complete this assignment?
FEEDBACK_NUM_HOURS: 21
# Please list any typos / errors you noticed in the assignment description or skeleton code
FEEDBACK_ERRORS: |
None
# Please describe any problems you encountered while completing this assignment
FEEDBACK_PROBLEMS: |
None