A twitter bot that uses machine learning to define invented words, posting truncated definitions on Twitter and complete ones on Tumblr. Tweet @lexiconjure a made-up word, and it'll define it for you. [more inside]
Neuralsnap generates an image caption using a model I trained (convolutional and recurrent neural networks), then uses another character-level recurrent neural net that I trained on ~40 MB of poetry to expand the caption into a poem. (In this example, generated from a Rothko painting, the red text is the direct image caption, and the rest is the poetic expansion.) [more inside]
'Alphabet Aerobics' is a pretty interesting track. I made it decidedly less interesting using some python signal processing libraries to make the whole track a constant tempo. [more inside]
My own humble contribution to the learn-to-program literature.
With Clarifai, an image concept extraction API utilizing convolutional neural networks, and ConceptNet, a lexical relationship database, I built a template system to generate paragraphs of text from photographs. word.camera is responsive — it works on desktop, tablet, and mobile devices. The code behind it is open source and available on GitHub, because lexography is for everyone. [more inside]
Opinion polls are all well and good, but they don't give you much of an idea of what might actually happen in an election (particularly in a multi-party democracy like the UK). Electobot aims to solve that by running thousands of simulated elections in order to work out what might happen if the election were run tomorrow with the polls as they are. In addition to running the simulations, I've also been blogging the results at Electobot: The Blog. [more inside]
Using Python 2.7 and the Natural Language Toolkit, I created a program called Sonnetizer that generates 14-line rhyming sonnets in (mostly) iambic pentameter from any text corpus. Using Sonnetizer, I generated 10,000 unique sonnets from the sonnets of William Shakespeare, and compiled them into a PDF.
I wrote a little library for working with functors, applicative functors and monads in Python. You can grab it from the Python Package Index. [more inside]
In the course of the recent end-of-year ritual of going through my past year's unfinished Ableton Live projects, I found myself writing a Python script to partially automate the task, by examining the Live sets and dumping information about them (namely, the tracks and the instrument/effect devices in them) as text. [more inside]
A course in programming in Python for literate non-programmers, offered in Brooklyn, NY. [more inside]
I have started a new technical blog, where I will be documenting various projects I have worked on (typically involving code, though not always). [more inside]
Anagramatron searches for anagrammatical pairs of tweets, the cream of which are retweeted and posted to an associated tumblr. Source on github.
GRARchiver is a small python script that archives all your Google Reader read items. You need to know how to use python to use it. [more inside]
Metropho.rs is a geographic metaphor map that plots "X is the Y of Z" tweets by putting the "Y" label on the "X" location. Some nice coverage by the Atlantic Cities blog here. [more inside]
Spotmaps is an on-going project to map the colour footprints of different films. The website was designed to show off the completed image library. [more inside]
We started out as Rails developers trying to script capistrano deployments and somehow ended up hacking together a full platform-as-a-service system, run on dedicated servers, with S3-compatible distributed storage and a REST API/build process that's mostly heroku-compatible (i.e. Procfiles and buildpacks and one-time-processes). We've got support for Ruby 1.9.3 and 2.0 preview including Rails, node.js/CoffeeScript, python, php, etc etc. [more inside]
A general-purpose dataflow programming language based on Python, written in Python [more inside]
Kicksaver finds Kickstarter projects that will end soon without meeting their funding goals. If you have a few dollars to spare, you can help tip a struggling project over its goal and prevent it from losing the pledges made so far. It also tweets endangered projects a few times a day. This is a weekend project I made as part of my continuing effort to teach myself programming. It's completely noncommercial, open source, and not affiliated with Kickstarter in any way (although I do think they're pretty great). [more inside]
My friends and I built an 80s Joke Line for a hackathon event here in Toronto. You can call in to hear a joke and then share one of your own. It's cobbled together using Flask (a Python web framework) and APIs from Twilio and Soundcloud. The code is on GitHub. We are a bit short on jokes. Maybe you can help with that. [more inside]
Ever wanted to know where in Oregon the next new bar will be? Every week, Oregon's liquor control commission posts pdf's of their applicants for a liquor license. This bot grabs the latest PDF, makes a google map from it, and then tweets it. [more inside]
This is a static blog compiler that works well with Git. [more inside]
Here is my video offering solutions to the global food crisis as part of a submission for the international Thought For Food challenge. Possible NSFW if your boss can't abide a male nipple. [more inside]
It's never too soon for more Markov chains, right? cobe is a Markov based text generation framework that behaves like the venerable MegaHAL chatbot, backed with a more modern (SQLite) datastore. Its brain is kept entirely on disk, uses little memory, and scales reliably. A very minimal IRC client is included. [more inside]