Software

Here are some computer things that I have made. Mostly experiments in machine learning, viz & co.

• smöl

  Analytically sumarize a .csv file right there in your browser, because small data matters too. Actually runs on pretty big data, thanks to duckdb-wasm.

• THE EMOTINOMICON

  An app for internal navigation.

• Life with Ghosts

  Conway's Game of Life, but dead cells decay instead of immediately disappearing.

• Binary Metrics

  An app for visualizing binary classification metrics as a function of the classification threshold.

• Bars

  A small horizontal bar-chart maker. Things are editable, get clicking or tapping.

• Fuzzy People

  A small Observable notebook fuzzing people out of images.

• Shakespair

  Have a Shakespearean character-LSTM complete your sentence. From before GPT ate text generation.

• Squote

  Semantic search using BERT embeddings, applied to questionably famous quotes.

• NBSVM

  A custom scikit-learn classifier to provide strong text classification baselines. Developed as part of Fast Forward Labs' research into transfer learning. A sample of that reasearch is at Textflix: Using Transfer Learning for a Natural Language Processing Prototype.

• Bayes' Theorem For Disease

  An interactive notebook introducing conditional probability and Bayes' Theorem in the context of testing for disease.

• Berkson's Paradox

  An interactive notebook demonstrating how spurious correlations arise from selection bias. Part of Fast Forward Labs' research on Causality for Machine Learning.

• An Invitation to Active Learning

  An interactive notebook introducing active learning by training two simple logistic regression algorithms live in the browser. Part of Fast Forward Labs' research on Learning with Limited Labeled Data.