Check our our final showcase video below!!

Overview of project:
* DQN, NFSP, and CFR using RLCard (open source Poker engine)
* PyPokerEngine GUI and final webapp

The Illustrator project is organized into 3 parts...
Sketch Classification:
* Given a snapshot of a drawing, what is the most likely word that the user is drawing?
* MobileNet trained over 20 classes
Sketch Generation:
* Given a word, non-deterministically generate a sketch of that word
* Implementation of Sketch-RNN based on this paper.
Application Architecture:
* The web infrastructure technologies holding up the application and game server
Technologies Used:
Python-socketio
* Allows for communication in two directions (server <-> client)
* Rooms for games
Aiohttp/Asyncio
* Enables asynchronous socket handlers and coroutines between client and server
* Switched to this after having a problem with handling synchronous handlers
React front-end

Our goal for the project was to replicate the results of the Facebook research, A Universal Music Translation Network, by rebuilding the network and running further experiments. The network has an encoder that encodes inputs into a representation that, theoretically, doesn’t include any instrumentation details (“domain specific information”). A decoder can convert the encoder output into an audio file with a specific instrumentation. Each instrumentation (“domain”) has its own decoder. However, the whole network is pretty computationally expensive, so we have experimented with:

• Speeding up training - using different inputs, testing different complexities, hyperparameters
• Creating a music classifier and comparing with the latent vectors to see if any insight can be gained there

The goal of the project was to build a network that would be able to answer a question about a given image. Our research focused mostly on working with computer vision and natural language processing, finding the best word embeddings for our questions (GloVe, Word2Vec, RNNs), and figuring out how to implement our own FCNN and attention models.

DataSelfie: Empowering People to Design Personalized Visuals to Represent Their Data
Published in ACM Conference on Human Factors in Computing Systems (CHI), 2019
Nam Wook Kim, Hyejin Im, Nathalie Henry Riche, Alicia Wang, Krzysztof Z. Gajos, Hanspeter Pfister
Abstract of the paper: Many personal informatics systems allow people to collect and manage personal data and reflect more deeply about themselves. However, these tools rarely offer ways to customize how the data is visualized. In this work, we investigate the question of how to enable people to determine the representation of their data. We analyzed the Dear Data project to gain insights into the design elements of personal visualizations. We developed DataSelfie, a novel system that allows individuals to gather personal data and design custom visuals to represent the collected data. We conducted a user study to evaluate the usability of the system as well as its potential for individual and collaborative sensemaking of the data.