Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
3 views

Tutorial1_ML_Cyber4

Machine learning

Uploaded by

wassimna0
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
3 views

Tutorial1_ML_Cyber4

Machine learning

Uploaded by

wassimna0
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Tuto1/AI 2024/2025

Tutorial 1: AI & Supervised Learning


Machine Learning Algorithms (CH1&2)

Exercise 1 : Machine Learning Jeopardy

Form 3 groups.
Choose a category from 1 to 3; (cat1: ML basics/ cat2: Supervised Learning/
cat3: ML Applications)
Choose a point value; (from 100 to 500 points)
Exercise 2 : Data preprocessing

1. What is data preprocessing in machine learning?


2. What is data mining?
3. Enumerate data preprocessing methods and explain each one giving a real
example of datasets.

Exercise 3 : Supervised vs Unsupervised ML

Classify these applications into supervised or unsupervised machine learning


applications.

1. Email Spam Detection


2. Customer Segmentation Based on Purchasing Behavior
3. Image Recognition for Facial Recognition
4. Predicting Housing Prices Based on Features (e.g., size, location)
5. Stock Price Prediction Using Historical Data
6. Recommending Movies Based on User Ratings
7. Clustering Customers into Groups Based on Similarities
8. Handwritten Digit Recognition (e.g., from scanned documents)
9. Time Series Forecasting for Weather Predictions

1 Supervisor : Yousra Hadj Hassen


Cyber4_AI
Tuto1/AI 2024/2025

Exercise 4 : Supervised ML: The classifier KNN

Assume that we have two classes, A and B and a new document d to be


classified. The following training data is available:

Let us assume that we use the cosine as a distance measure, i.e., the higher the
cosine, the closer are two vectors.

Which class would be assigned to d with a k-nearest neighbor classifier using


cosine if;
1. k = 3 and simple majority vote
2. k = 5 and simple majority vote;
Exercise 5 : Supervised ML: The classifier Support Vector Machine

1. Consider a linear SVM with decision boundary g(x) = wTx+w0. In SVM


classification, explain why it is useful to assign class labels -1 and 1 for a
binary classification problem.
2. The basic SVM optimization problem is to minimize J = ½ ||w||2 What are
the additional constraints for this optimization problem? Ideally, you
should answer both by math and explain what this expression means.
3. Explain how slack variables i are used to solve a non-separable case like
the one below:

2 Supervisor : Yousra Hadj Hassen


Cyber4_AI
Tuto1/AI 2024/2025

4. Discuss how likely a Gaussian classifier and an SVM classifier are to


overfit to the training data.
5. Explain how an SVM can be used on a classification problem with M
classes.
6. Explain briefly how SVM parameters should be determined.

3 Supervisor : Yousra Hadj Hassen


Cyber4_AI

You might also like