Jasmine S
- Research Program Mentor
PhD candidate at Columbia University
Expertise
computational neuroscience, computational modelling, learning in artificial neural networks
Bio
A Neurobiology and Behavior Ph.D. student with Ashok Litwin-Kumar at Columbia University, Jasmine is interested in investigating biological algorithms for learning using computational models. She is also interested in developing new machine learning techniques using insights from this work. She began doing neuroscience research as a high schooler and is excited to mentor high schoolers in research. Previously she developed and released PsychRNN, a software package for modeling decision-making, memory, and other cognitive tasks using recurrent neural networks, and developed methods for analyzing neural data. Jasmine grew up in Austin, Texas, where she enjoyed synchronized swimming, social swing dancing, and playing violin. Since then she has played water polo and kayak polo, but now mostly swims. She is still swing dancing socially and playing violin. She also sometimes enjoys making art badly--pottery, painting, drawing, whatever is accessible.Project ideas
Learning in a Recurrent Neural Network
With this project you will train a recurrent neural network to do a task of your choosing -- maybe one that animals or people are trained to do. What aspects of training affect how well the network learns or how it solves the task? With this project you will strengthen your Python and Machine Learning skills, and you can delve into the neurobiology of learning as much or as little as you want. You could create a scientific paper, blog post, presentation, or video about your work. You could also explore how biological constraints on neural networks affect learning and task performance, or work with different learning algorithms.
Analyzing the Fly Connectome
With this project you will make use of the publicly available fly connectome which maps how each neuron in the fly brain connects to every other neuron in the hemibrain dataset. You can make predictions about how different brain regions talk to each other. We would start by reading about what different brain regions in the fly do and pick some areas of interest -- for example, how does the learning system talk to regions of the brain that control movement? You will learn how to work with matrices in Python, plot data to visualize it, and read scientific papers. You could create a scientific paper, blog post, presentation, or video about your work.