ELEC 400M
ELEC 400M : Machine Learning Fundamentals for Engineers.
| My Class Average: | 88% |
| My Grade: | 93% |
| historical averages: | 84.07% |
| https://ubcgrades.com/statistics-by-course#UBCV-ELEC-400M | Â |
My Impressions:
Interesting course, but it was too math heavy and lacked good explanations for machine learning concepts. I often found myself explaining the purpose behind methods to my friends because the class lecture was focussed on why it mathematically made sense and not logically.
Homework
ALL my homework files can be found here https://github.com/amunwes/Elec-400M in jupyter notebook format, I suggest running them on google colab.
HW1
is a nice tutorial on Logistic regression in depth.
HW2
illustrates an example of using Principle Component Analysis to reduce the complexity of a classification problem and then runs a classification using Linear SVC, SVC using an RBF kernel, and RandomForrest. Comparing the results obtained at the end.
HW3 & HW4
work through examples of building a pipeline and building and utilizing Neural Networks in pytorch. They both show comparisons of accuracy metrics as you alter the number of epochs the networks are trained for and how to optimize the pipeline.
Final_Project
is my Final term project for the course. The project was fully independant and I chose to do a natural language processing classification project. I took a database of 40,000 movies and their synopsis’ and cleaned the dataset using natural language processing to remove redundant words and tested how 3 classification algorithms manage the task of multi-genre classification using the synopsis word counts as features. In the end I wrote a comprehensive report found here detailing my process and findings.