Assignment 1: Training MNIST + Project Declaration
Due 02/11/2024, noon AOE
Credit: 5 points
This project has two parts. In the first, we will brush up on concepts learned in the tutorial through a video lecture from youtube, and in the second you will be declaring your project following the things learned from this video.
Learning outcomes:
(1) Getting comfortable with the machine learning "pipeline" i.e. the steps involved in a machine learning experiment. In the second assignment, we will be delving deeper into the different parts. So it is imperative that you get comfortable with these steps through the tutorial and this assignment. (2) With the above steps in mind, thinking through your project. How would these steps pan out in the case of your own class project?
Rubric:
1. 2.5 points for a working MNIST example.
2. 2.5 points for the project declaration report.
Deliverables:
Part A: MNIST Exampl
1. If you're using Google Colab, Pytorch comes pre-installed. If you're using your own system or machine, you will need to install pytorch (https://pytorch.org/get-started/locally/), and maybe some other python packages like Numpy etc.
2. Watch this video to completion:
https://www.youtube.com/watch?v=OMDn66kM9Qc (Links to an external site.)
3. Follow the different steps discussed here, and complete the MNIST example in PyTorch as shown there.
4. Submit a screen-grab video running your written code. It is strongly suggested to not copy the code. Following the tutorial and writing it along will greatly help understand the fundamentals. For full credit, your video should show your written code running to completion without any errors.
Part B: Project Declaration
Based on your understanding of a machine learning pipeline as described in the video in Part A above (i.e. dataset, model input, model output, model architecture, evaluation metrics, etc) report your project idea in about a 1 page PDF. Pick whatever is easiest for you: Word -> PDF, Google Docs -> PDF, or LaTeX -> PDF.
The goal is to listing each step in a concrete, concise manner. This will help us identify the less concrete points in your project, and help you with them early on. Below is a sample project declaration for a dummy project which could be used as a template.
Sample Project Declaration - Neuro 140_240.pdfPart C: Indication of Past Experience (Optional and not graded)
Students enter this course with varying degrees of past experience with machine learning, and we will take this into account when we are grading the final projects. If you wish to tell us anything about your past experience, you may do so here in 1-2 sentences.
For example: "I know how to write code in MATLAB, but I have no past experience in machine learning", "I have taken a basic machine learning course, but haven't done a significant ML research project before", "I have published machine learning papers in computer vision, but have no experience with natural language processing. I am excited to expand my expertise by doing an NLP-focused project". This is completely optional and does not count towards the grade. If you don't complete this, we won't make assumptions about your background one way or the other.
This is NOT intended to penalize folks who come in with lots of knowledge and experience or students without ML expertise - regardless of your background, we hope you'll dive into an exciting and challenging machine learning project this semester.
