I believe that good teaching cannot happen without student trust in the design and purpose of a course. To build up this trust, it is imperative to respect that students come to a class as entire people from diverse backgrounds and circumstances. Their time is finite, so good teaching must be rigorous without being e-n-d-l-e-s-s. Their input is valuable, so good teaching cannot be so rigid that it does not account for differences among learners and classes. Their motivation is earned, and good teaching takes strides to earn as much of it as possible from each student.
I am committed to teaching practices that include diverse identities and perspectives not simply because these practices are fair and socially just, but because they are fundamentally good teaching practices. When a course is designed with inclusivity in mind, an instructor & the students construct an environment wherein all participants will be encouraged to learn and grow.
And I am committed to continually improving.
An experiment in iteration.
Top Tuneage from 2010-2019
Are you better at diagnosing rare diseases than most American doctors?
I collected the data for this Pudding project. Thanks to TripAdvisor for being so patient with all the web-crawling.
Data viz and analysis by author Ilia Blinderman.
What factors determine an author's voice in writing? We used state of the art NLP techniques to define features that can be used to quantify the notion of authorial style. In particular, we explored how constructs like passive voice usage and entity coreference patterns create an author's signature literary fingerprint. Our data came from the Guardian's opinion pieces and book reviews over the past decade, so this particular experiment was performed in the domain of non-fiction. Co-authored with Samuel Sharpe, Justin Cho.
The ability for computers to create and understand human-readable source code is an attractive goal with profound implications for software engineering and security research. In this paper, I provide a survey of research which has attempted to generate or learn programs using Neural Networks and evaluate strengths and weaknesses of each. Additionally, I formalize an extention to the promising work done by DeepCoder that incorporates a Graph Neural Network into the inductive framework.