Chroma is a patent pending wearable real-time augmented-reality application that promises to address real-life issues of color-blind people. Chroma transforms the colors of the scene into new colors depending on what the user wants to see. I started Chroma with Google Glass as the target platform. I am currently developing a better algorithm for Chroma and the iOS version of Chroma.
CortexSuite is a Synthetic Brain Benchmark Suite which captures an emerging workload that is clustered around providing information processing capabilities that human brains have today but that computers are only just now beginning to become good at.
The goal was to facilitate the analysis, transformation, parallelization, and simulation of the benchmarks by
compiler and architecture researchers.
I implemented the benchmarks for movie suggestion (using Restricted Boltzmann Machine), speech recognition (using CMU Sphinx), and singular value decomposition.
I worked with a PhD student under Dr. Kamalika Chaudhuri to formulate a new linear classifier with reject option, derived from Minimax Probability Machine. In doing so, we tried several different methods to train the classifier, compared the results, and analyzed the algorithms to understand their behavior in different datasets.