Assignment 1
Perceptron
Out: 09/01/2026. Due: before midterm break.
Consider the following training set (gzipped) on 1024 dimensions:
train.
Train a Boolean linear classifier (using the Perceptron algorithm) on the training set.
What to submit
Submit the following via email:
- The classifier (as a real-valued normal vector for the line separator).
Use the same format as a single row in the training file (without the label).
- The margin parameter achieved by your linear classifier on the training sample.
Simply include this in your email.
- Source code.
Provide your own implementation (without using built-in functions from ML packages).
You may work in groups of two.
AI usage:
You may choose to use the assistance of AI tools for this assignment. But, in that case, you should include a report that documents this. State which tools were used, how did the AI tool assist (or not) in this assignment, and an assessment on the experience (in relation to not using the AI tool).