Adithya Ganesh, Allan Jiang
Developed curriculum and lectures for course focusing on functional programming in Clojure. Topics include: immutability, functional programming (function composition, higher order functions), concurrency (atoms, promises, futures, actors, Software Transactional Memory), LISP (REPL-driven development, homoiconicity, macros), and interop (between Clojure code and code native to the host VM).
Adithya Ganesh, Joe Charalel, Matthew Das Sarma, Nancy Xu
Applied deep reinforcement learning to train an autonomous driving system in the TORCS simulator. Deep feedforward neural networks were implemented in TensorFlow / Keras, and the algorithm demonstrated robustness over a diverse range of track geometries.
Deep convolutional neural networks for diabetic retinopathy diagnosis.
Developed system capable of diagnosing and assessing severity of diabetic retinopathy from patient retinal images. Deep convolutional neural networks were implemented using the Theano and Lasagne frameworks, and were deployed on AWS GPU instances. Training dataset consisted of ~85,000 high-resolution retinal images from eyePACS and the California Healthcare Foundation.