Lecture 05

Linear models, logistic regression

Important

Deadline: Assignment 1 must be submitted no later than September 29, 2024, at 11 PM. Please refer to the assignment description available on Brightspace.

Prepare

• Decision bounddary: Russell and Norvig (2020), pages 658, 682
• Logistic regression: Russell and Norvig (2020), pages 684-686
• Watch Machine Learning and Logistic Regression, IBM Technology.

Participate

• slides

Practice

In class, we developed a logistic regression model for handwritten digit recognition using a dataset from UCI ML. This dataset comprises 1797 images of size \(8 \times 8\). The MNIST (mnist_784) dataset contains 70,000 images of size \(28 \times 28\). The following example, from the sklearn website, uses this dataset and graphically presents the coefficients (\(\theta\)) for each of the 10 models. You can load this model as follows:

from sklearn.datasets import fetch_openml  

X, y = fetch_openml("mnist_784", version=1, return_X_y=True, as_frame=False)

References

Russell, Stuart, and Peter Norvig. 2020. Artificial Intelligence: A Modern Approach. 4th ed. Pearson. http://aima.cs.berkeley.edu/.