Skip to content

Latest commit

 

History

History

hw2_classification

README.md

HW 2 - Classification

This is official repository for the second homework assignment of the course UROB. The assignment is about classification - its understanding and implementation.

Sections

What will you learn?

In this homework, you will start by reviewing the basics of classification, and then you will implement a simple classifier. Then you will learn about more advanced classifiers using deep learning. After this homework, you will have a broad understanding of classification, and you will understand these terms:

  • Cross-Validation: A Vital Model Assessment Technique

Cross-validation emerges as a crucial technique for evaluating machine learning models. It operates by training multiple ML models on subsets of the available input data and subsequently evaluating their performance on the complementary subsets. While this method can be computationally demanding, it yields substantial benefits. By employing cross-validation, you can maximize your utilization of the available data for model assessment, a particularly advantageous approach when working with limited data resources.

  • Linear Classifier: The Fundamental Building Block

Linear classifiers constitute a foundational category of algorithms within the machine learning landscape. These algorithms play a pivotal role in assigning input values to discrete categories. While they represent one of the simplest classification algorithms in the field, they serve as the cornerstone for numerous other advanced classification techniques.

You will explore the linear classifier from two distinct angles:

Geometric Perspective



Template Matching Perspective

  • Neural Network Classifier: Unveiling Complexity

Among the most intricate classifiers you will tackle is the neural network classifier. This journey will demystify neural networks and unveil the inner workings behind their decision-making processes.

Complex Decision Boundaries of Neural Networks

During your exploration, you will experiment with various activation functions and observe their effects on decision boundaries.

Activation Functions

Furthermore, you will witness how neural networks learn features by transforming input data into a novel space. This journey also highlights the connections between neural networks and linear classifiers.

Feature Space Transformation

By delving into these facets, you will gain profound insights into the intricacies of neural network classifiers, enhancing your comprehension of this powerful machine learning tool.

Getting started

To begin your journey with this assignment, follow these simple steps:

  1. Clone the Repository: Start by cloning this repository to your local computer. Use the following command in your terminal:

    git clone https://github.com/urob-ctu/hw2.git

  2. Install Dependencies: We recommend using a conda environment to manage dependencies. Create the environment by executing:

    conda env create --name hw2 --file environment.yaml

    After the environment is created, activate it with:

    conda activate hw2

    If you are using a different environment manager, please refer to the documentation for the appropriate commands and make sure you have Python 3.10 installed.

  3. Choose Your IDE: Next, open the project in your preferred Integrated Development Environment (IDE). While we suggest PyCharm, the choice ultimately lies with you.

    Warning If you are using PyCharm, you need to disable the invert image outputs for dark themes if you are using one. You can do it by going to Preferences | Languages & Frameworks | Jupyter | Invert image outputs for dark themes. For users of alternative Integrated Development Environments (IDEs), please locate the analogous setting. Failing to make this adjustment could result in images that might not look as visually appealing as intended. :)

Working on the Assignments

The homework is divided into several sections, each residing in its own notebook. You can locate these notebooks in the project's root directory. To make steady progress, adhere to the following sequence:

  1. Introduction to k-Nearest Neighbors
  2. k-Nearest Neighbors: Hyperparameter Optimization
  3. Introduction to Linear Classifiers
  4. Linear Classifier as a Template Matching Algorithm
  5. Introduction to Neural Networks
  6. Training the Multilayer Perceptron

Inside each notebook, you will find task descriptions and the specific files you need to modify. These editable files are situated in the assignments directory. Please refrain from altering any other files. Within these designated files, make changes only to sections resembling the following:

# ▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱ Assignment 1.1 ▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰ #
# TODO:                                                             #
# Calculate the L2 distance between the ith test point and the jth  #
# training point and store the result in dists[i, j]. Avoid using   #
# loops over dimensions or np.linalg.norm().                        #
# ▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰▱▰ #
# 🌀 INCEPTION 🌀 (Your code begins its journey here. 🚀 Do not delete this line.)
#
#                    ╔═══════════════════════╗
#                    ║                       ║
#                    ║       YOUR CODE       ║
#                    ║                       ║
#                    ╚═══════════════════════╝
#
# 🌀 TERMINATION 🌀 (Your code reaches its end. 🏁 Do not delete this line.)

Remember, any modifications outside of this designated section can cause the autograder to fail, resulting in a suboptimal grade. If you have any questions, please reach out to the teaching assistants.

To test your code, you can run the following command in your terminal:

python test_assignments.py

How to submit

Once you've completed the assignment and are ready to submit your work, use the following command in your terminal:

python submit.py

This will create a zip file named hw2.zip in the project's root directory. Submit this file to the BRUTE system.