Welcome to Machine Learning homework 2!

Hi! I'm Edoardo and i'm a student of Sapienza University and this repository is the second assignments of Machine Learning course of Artificial Intelligence and Robotics.

Files

Here you can find the request of this assignment and the exercise generator that provide to choose the exercises starting from the MATRICOLA

Machine Learning Assignment

In this homework, you are invited to provide a solution for the classification of images regarding objects in a home environment.

Dataset and instructions are provided in the attached Colab.

To solve the problem, you can use any method, any tool, any programming language. It is your choice to pre-process the input images in any way it is useful for the method you are using.

The output of the assignment should be:

1. A report (PDF file of about 10 pages, with your name and matricola code) describing the implemented solution: how data have been preprocessed, which method/algorithm has been used, which configurations of the method have been tried, description of the evaluation method used, and obtained results using appropriate metrics. The report should compare at least two different solutions, obtained, for example, by changing the value of an hyper-parameter of the used method, by testing different ways of preprocessing, etc. Conclusions should discuss the comparative results. Computational training time can also be interesting to report and comment.
2. A ZIP file with the code you used in the project (do not include the dataset)
3. [OPTIONAL] If you are using Keras and Tensorflow, save your best trained model, using the function model.save(...), and submit it in a file named .h5 (e.g., 1234567.h5).

Submit the three files in this assignment. Make sure to turn the assignment in, otherwise it will not reach the teachers. NOTE: do not put the PDF report into the ZIP file!!!

This assignment must be individual (i.e., one submission for each student) and original (i.e., not equal or too similar to other works either from other students in this class or from other sources). Evaluation will be based on the appropriateness and correctness of the described solution, regardless of the numeric results (as long as they are reasonable). The results on the blind test also do not affect the evaluation of this homework.

Here you can find the generated model directory:

Generated Models (This link was removed, because it has a big size. If you want to see my result, please contact me!)

Solution

Here you can find the solution jupiter notebook:

Solution

Here you can find the PDF that describe the implemented solution:

PDF REPORT

Note: Remember to import che correct location of the file! Please upload all the folders and files into de Google Drive's folder and after that change the path inside the project with your one.