A handwritten digit classifier using built using a Convolutional Neural Network (CNN). Used various techniques such as data augmentation, batch normalisation, and dropout to improve the model's performance.

Comparing various classification algorithms to predict whether an adult earns more than $50,000 a year. Emphasis is on feature engineering, data preprocessing with One-Hot Encoding, and model optimization through hyperparameter tuning.

Built a neural network from scratch to teach a Pacman agent. Used various techniques such as batch normalisation, dropout, momentum, learning rate decay and more to improve performance.

Practicing various algorithms and techniques. This includes supervised, unsupervised, and reinforcement learning algorithms, as well as feature engineering, data preprocessing, and hyperparameter tuning.

Practicing various Reinforcement Learning algorithms and techniques. This includes Q-Learning, Deep Q-Learning, and Asynchronous Advantage Actor-Critic (A3C) algorithms.

A custom Q-Learning agent learns to play Pacman. Required foundational Mathematics knowledge and understanding of the Q-Learning algorithm.

Implemented various algorithms and techniques learnt during the course, such as Nearest Neighbours, Conformal Prediction, Regression algorithms, data preprocessing, kernels, Neural Networks, SVMs, etc.

Focusing on learning generative models, using third-party models and using advanced techniques. This includes techniques such as transfer learning, LLM Agents, and Generative Models.