Jake Liebow, Dingjun Weng, Albert Dayn
Project Tracker: Trello Video: YouTube Deployed App: TetrAIs (this isn't updated yet) Repo: GitHub
- /ai - Python files defining the AI and a python implementation of Tetris
- /*.old - original architecture attempt (Python with Flask, Elm)
- /web - Current Source code / Tests / Database setup
- /web/*.php - Middleware
- /web/src - Javascript sources
- /web/tests - Javascript tests
- /web/database - database setup scripts
- /web/styles - Frontend CSS
Just copy the entire web folder to a PHP compatible server directory, then access the app starting at index.php
We use the Jest unit testing framework, so to run tests you will need to install Jest with npm.
To do so, simply run npm install. To run the tests, run the command npm test within the /web directory.
We are going to make a website where users can play Tetris in their browsers. To provide a more authentic, retro experience, users will be asked for a name at the end of their play to put up on a global leaderboard. If time permits, we will add the ability to store speedruns of the game so players can see how others got to their high scores. For a more modern twist on this beloved game, we will use Tensor-flow to learn an AI to play Tetris at the same time as the user, so players can compete against a challenging AI in realtime. We call this Tetris-mastering algorithm: TetrAIs. It will play the same game as the player, with blocks falling at the same time for each oponent. TetrAIs will be the ultimate competitor to even the most skilled players, testing their resolve. With the AI always learning and getting better, we force the user to continuously improve if they want to keep up.
Here at TetrAIs we are commited to providing you with the most challenging Tetris opponent you have ever faced, a machine learning algorithm that evolves with every challenge it faces. TetrAIs is always getting better... Are you?
We wanted to use this project as an oportunity to learn about machine learning and tensor-flow, and to create an awesome, fun game in the process.
As with any deep learning project, it may take too long to train a useable AI to solve the problem of playing Tetris. Our team has a lack of experience with both front end development and machine learning using neural nets. We only have two people.
We will start on the AI as soon as possible to reduce the risk of not having enough time for the algorithm to converge on an optimal solution and to maximize the time we have to get familiar with the problem domain.
Github
We will be using the Agile methodology with 2 week long sprints. Since we won't be able to meet every day, we will be using Slack to post our status and ask any questions of each other in place of a daily Stand Up. We will meet once a week in person, with every 3rd meeting being a retrospective to decide how we can improve for the next sprint.
We have setup a Slack chanel for this project.
We will be using Google's tensorflow neural network library to learn an AI to play tetris, which will be done using Python. MySQL will be used for the backend database to store high scores. We will use Elm as our front end language and Elm-UI as the front end framework. This will be the implementation language for our tetris game and the runner of the neural network sent from the server. For the server, we will use Python and the Flask framework. We will use it's built in development server as opposed to adding on a standalone application like Apache or NginX.
Description: Our database for the A.I. powered gaming website is made up of two tables, 'users' and 'leaderboards'. The generic table 'users' which holds the player id(primary key, enforces uniqueness), username, hash encrypted password and high score of the player. The table 'leaderboards' tracks the top 10 high scores and their record holders, this table is made up of the rank(primary key enforces uniqueness), user_id ( foreign key to access specific player on leaderboards from users), leader_name, and the high score itself.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent