Machine Learning Projects

Computer Vision – Image Classification

September 2021

Python; pandas, TensorFlow

• Built and trained a convolutional neural network (CNN) to classify 60,000 images from the CIFAR-10 dataset into ten categories with 74.3% accuracy.
• Augmented the training samples by adding random rotations and translations, further improving the classification accuracy to 75.6%.

[Code]

Natural Language Processing (NLP) – Spam Classification

August 2021

Python; pandas, scikit-learn

• Transformed natural English words gathered from email messages into numerical vectors using count vectorization.
• Built a naïve Bayes classifier to distinguish spam emails from legitimate emails with 98.9% accuracy.

[Code]

Movie Recommendation System

July 2021

Python; pandas, NumPy

• Analyzed the correlations between ratings given to different movies using pandas and NumPy.
• Constructed an item-based collaborative filter from the correlation matrix to predict users’ ratings based on their watching and rating history.

[Code]

Recurrent Neural Network (RNN) Implementation – Time Series Prediction

April 2020

Python

• Implemented a RNN from scratch, optimized using gradient descent with backpropagation.
• Evaluated the RNN with a time series of a noisy sine wave, correctly predicting subsequent values.

[Code]