Generator is MIT’s premiere research competition intended to recognize promising students who are improving and harnessing artificial intelligence to solve societal challenges.

Through mentorship from industry and academia professionals, support for cloud compute resources, and access to our events and speaker series, Generator empowers undergraduate and graduate students across Boston-area universities on a variety of machine intelligence projects, encouraging technical and creative excellence and promoting a passion to solve the problems of the present and future.

Teams compete in several rounds leading up to the final demo day in May 2021, at which they will give (virtual) presentations on their projects and key results to compete for our top prizes. Judges are asked to take into consideration technical execution, product potential, creativity/novelty, and presentation skills when reviewing project presentations. The top three undergraduate and graduate teams will be awarded prizes ranging from $500 to $2K.

Past winning projects include:

1. Graph Transformer Neural Networks for Molecular Prediction (First Place, 2019)
   
   Applying a Graph Transformer Neural Network (GTNN), an MPNN variant that draws from NLP’s self-attention block, to the QM9 quantum chemical property dataset for molecular prediction.


2. Chains and Layers in Evolving Artificial Neural Networks (Second Place, 2019)
   
   A novel genetic algorithm, Chains and Layers in Evolving Artificial Neural Networks (CLEANN), which solves some of the limitations of other genetic algorithms and proves effective in both supervised and unsupervised learning problems.


3. Boston Agritech — Automated Indoor Farming (Third Place, 2019)
   
   A machine intelligence platform to help farmers automate critical tasks such as pest monitoring and growth monitoring using deep convolutional neural networks, segmentation, and linear regression.

Generator would like to thank all of our amazing sponsors for supporting our continued efforts:

Generator was started in 2018 by the Artificial Intelligence Club at MIT (AIM). If you have any questions, please reach out to the Generator coordinators at aim-generator [at] mit.edu.