FrickFrackFractal

The name says it all, no?

  • An interactive fractal generator coded in Java
  • Fractal generation through the Iterated Function Systems (IFS) method with GUI to allow more freedom in program

GitHub Link: https://github.com/shraz12/FrickFrackFractal

Bayesian MNL Regression

As if the phrase multinomial logit regression wasn’t long enough– now we’re adding ‘Bayesian’ to it.

  • A model for Bayesian Multinomial Logit Regression following the work of Leonhard Held and Chris Holmes (see citation below)
  • Application to prediction problems in NLP in conjunction with a deep neural network used to embed words into vector space

Leonhard Held. Chris C. Holmes. “Bayesian auxiliary variable models for binary and multinomial regression.” Bayesian Anal. 1 (1) 145 – 168, March 2006. https://doi.org/10.1214/06-BA105

GitHub Link: https://github.com/shraz12/BayesianMNLRegression

Generative Approaches to Sleep Pattern Inference

A sleep study certainly seems useful for college students… let’s hope they’ll actually listen when we say they need more sleep.

  • Set of models coded in Python used to infer sleep patterns in college students from fitness tracking data
  • Use of logistic regression, convolutional neural networks, and Hidden Markov Models to approach problems in time series segmentation
  • Work done in collaboration with Dichuan Gao and Alexander Young of Brown University
  • Winning paper in the Brown Mathematical Contest in Modeling (Brown MCM)

GitHub Link: https://github.com/shraz12/Generative-Sleep-Models

Reinforcement Learning In Forest management

Who would have thought that cutting down trees might actually improve Carbon Dioxide content in the air?

  • Set of models coded in Python to approach problems forest management via carbon sequestration
  • Use of dynamical systems models paired with Deep Q neural network to optimize tree cutting policies
  • Work done in collaboration with Qiuhong Anna Wei and Alexander Young of Brown University

Report Link: https://shreyasrao.net/wp-content/uploads/2023/06/ICM_tex-1.pdf

DeepWIFF

I promise this has to do with ice floe fractures, not deep sniffs of suspicious odors.

  • Pair of deep neural networks coded with Python’s TensorFlow library used to simulate the fracture of ice sheets, an important problem in climate science
  • Work done in collaboration with Professor Chris Horvat of the Institute at Brown for Environment and Society

GitHub Link: https://github.com/shraz12/deepwiff