Selected For GSoC 2019

Opinions, Experience and Expectations Ahead

Finally, after a nervous month of anticipation, GSoC results are finally out, and guess what, I have been selected. Here starts the journey of a summer long session of coding and a great oppurtunity to contribute to the open source community.

selection My project is selected under The Julia Language in Flux.jl. Flux is basically the machine learning spearhead of julia.

I will be working on making generative models such as cycleGAN and pix2pix for multi-modal image to image translation. These will be trained and made available for everyone to use as a part of Metalhead.jl. The next phase will involve reinforcement learning algorithms for continuous control like Trust Region Policy Optimization and Proximal Policy Optimization. These algorithms have a great baseline implementation in python and are state of the art in their applications. Most researchers thus resort to using these baseline implementations. Adding these to model zoo would hopefully attract a part of the reinforcement learning community to using Flux for research purposes. The final phase would involve Image Captioning Networks and Super Resolution GAN for single image super resolution.

Looking forward to a great time in the upcoming summers!