jalovisko/Classification-of-Image-Data

The 3rd mini-project (Classification of Image Data) of COMP 551 - Applied Machine Learning

Mini-project 3 — Classification of Image Data

The third mini-project of COMP 551 - Applied Machine Learning.

Useful links

• The project description is available here: http://www.reirab.com/Teaching/AML20/551P3.pdf

Repository description

Image classification is performed with multi-layer perceptron (MLP), convolutional neural network (CNN), and an adjusted multi-layer perceptron (AMLP) which was developed for this work.

Below are the files that cover the whole project.

• COMP_551_MLP_Image_Classification.ipynb — MLP code
• comp_551_mlp_image_classification.py — the same code for MLP but not as a Jupyter notebook, serves as an alternative to the Jupyter Notebook (but Jupyter Notebook has higher interactivity and thus recommended for use).
• MiniProject3_2_Keras.ipynb — CNN code

Feedback

• Completeness (20 out of 20): Everything was implemented.
• Correctness (38 out of 40): The CNN accuracy is a bit low.
• Writing (25 out of 25):
• The writing is good.
• Originality (15 out of 15):
• Good additional experiments.

Total: 98%

Note

The repositary was private for the duration of developement and was made public on May 9, 2020 (Happy Victory Day!).