build a simple AI using constraint propagation and search to solve Sudoku puzzles

build a game-playing AI agent to win the board game Isolation

use logic, planning and search to implement a system that efficiently moves cargo from their origins to their destinations using the least number of flights

use Hidden Markov Models to translate sign language into their English language characters

build a Pac-Man AI that finds the most efficient path through its world

• P1 - Search: implement depth-first, breadth-first, uniform cost, and A* search algorithms. These algorithms are used to solve navigation and traveling salesman problems in the Pacman world.

• P2 - Multi-Agent Search: implement multiagent minimax and expectimax algorithms, as well as designing evaluation functions.

• P3 - Reinforcement Learning: implement model-based and model-free reinforcement learning algorithms, applied to the AIMA textbook's Gridworld, Pacman, and a simulated crawling robot.

• P4 - Ghostbusters: implement exact inference using the forward algorithm and approximate inference via particle filters.

• P5 - Classification: implement standard machine learning classification algorithms using Naive Bayes, Perceptron, and MIRA models to classify digits then extend this by implementing a behavioral cloning Pacman agent.

• P6 - Contest: Pacman Capture the Flag: create strategies for a team of two agents to play a multi-player capture-the-flag variant of Pacman.

implement simulated annealing to solve the Traveling Salesman Problem (TSP) between US state capitals

use Backtracking to solve the N-queens problem (which is a generalization 8-queens problem)

build a pipeline that can be used within a web or mobile app to process real-world, user-supplied images. Given an image of a dog, the algorithm will identify an estimate of the canine’s breed. If supplied an image of a human, the code will identify the resembling dog breed.

build RNNs that can generate sequences based on input data - with a focus on two applications: With the first we'll use real market data in order to predict future Apple stock prices using an RNN model. The second one will be trained on Sir Arthur Conan Doyle's classic novel Sherlock Holmes and will generate wacky sentences based on it that may - or may not - become the next great Sherlock Holmes novel!

use IBM Watson's NLP Services to create a simple question-answering system

build a deep neural network that functions as part of an end-to-end machine translation pipeline

build a fully functional skill for Amazon’s Alexa that provides year-dated facts from AI History use the Alexa Skills Kit (ASK) - a current state of the art API for building voice systems.

track faces in a video, identify facial expressions (using Affectiva’s Emotion-as-a-Service API), and tag each face with an appropriate emoji next to it. Then turn this into a game where the player needs to mimic a random emoji displayed by the computer!