This document discusses a study on detecting cyberbullying using machine learning and deep learning techniques. Specifically, it examines using a hybrid model combining K-Nearest Neighbors, Support Vector Machine, and Random Forest algorithms, as well as a Convolutional Neural Network. The study uses a Twitter dataset to classify tweets as not bullying, racism, or sexism. It finds that the CNN model produces more accurate predictions than the hybrid stacking algorithm. The document provides background on related work applying machine and deep learning to cyberbullying detection, particularly using content-based and user-based approaches.