Tag Archives: open source

Human augmentation through artificial intelligence

Last week, I published two posts that relate to a very interesting and socioeconomically relevant machine learning topic: human augmentation through artificial intelligence. There have been many examples that made their way in the mainstream media where we’ve seen AI push the limits of what was possible, especially since deep learning really took off. One of the big recent examples is of course, AlphaGo.

But human augmentation isn’t about what AI can do by itself. Rather, it’s about how AI can be used to assist a human so that he or she can be much more efficient at performing a given task. It can be to make humans faster at it, or allow them to produce work of higher quality. It’s about using AI as a tool, just like any one of the hundreds of tools we all use in our everyday life.

The first post was a guest post on KDNuggets (with a companion notebook) on MLDB, showing how deep and transfer learning can be used to quickly train a model to classify if the car in a picture is a Tesla, a BMW or an Audi. It lays some of the technical foundation used in the second post.

The second post is about DeepTeach, the interactive Deep Image Classifier Builder. DeepTeach is an MLDB plugin, that is open source, and implements a human augmentation workflow allowing a user to quickly build an image classifier from unlabelled data within minutes.

As the post explains, the plugin uses a series of machine learning techniques, like deep, transfer and active learning. But what’s the most interesting about it is the workflow it implements, allowing a user to go from thousands of unlabelled images to a working binary image classifier in a minute. The same workflow could be implemented to deal with any type of data, like sound or text, and it concretely shows how AI can be used to make humans more productive.

The video below shows a demo of the plugin in action:

 

64-bit Scientific Python on Windows

Getting a 64-bit installation of Python with scientific packages on our dear Windows isn’t as simple as running an apt-get or port command. There is an official 64-bit Python build available but extensions like numpy, scipy or matplotlib only have official 32-bit builds. There are commercial distributions such as Enthought that offer all the packages built in 64-bit but at around 200$ per license, this was not an option for me.

Stumbled upon the Python Extension Packages for Windows page that contains dozens of extensions compiled for Python 2.5, 2.6 and 2.7 in 32 and 64 bits. With these packages, I was able to get a working installation in no time.

My time at Sun Labs and pyaura

My internship at Sun Microsystems Labs, which has been going on for about 15 months – 9 of those full time at their campus in the Boston area – is coming to an end. During the course of those months, I’ve met a lot of very smart and fun people, I’ve worked on very challenging and stimulating problems and I’ve discovered a bunch of really good New England beers.

All my work has been centered around the Aura datastore, an open-source, scalable and distributed recommendation platform. The datastore is designed to handle millions of users and items and can generate content-based recommendations based on each item’s aura (aka tag cloud).

Last summer, under the supervision of Paul Lamere, I worked a lot more on our music recommendation web application, called the Music Explaura and designed a steerable recommendation interface. (We also have a Facebook companion app to the Explaura that was created by Jeff Alexander.)

This summer, I worked with Steve Green on many different things, including what I’d like to talk about in this post, pyaura, a Python interface to the datastore.

pyaura

The idea behind pyaura is to get the best of both world. While the datastore is very good at what it does – storing millions of items and being able to compute similarity between all of them very quickly – the Java framework surrounding it is a bit too rigid to quickly hack random research code on top of it. While my actual goal was to experiment with ways of doing automatic cleanup and clustering of social tags, I felt I was missing the flexibility I wanted and was used to getting when working on projects using Python’s interactive environment.

Without going into details, since the datastore is distributed and has many different components, it uses a technology called Jini to automatically hook them all up together. Jini takes care of automatic service discovery so you don’t have to manually specify IP adresses and so on. It also allows you to publicly export functions that remote components can call. A concrete example would be the datastore head component allowing the web server component to call it’s getSimilarity() function on two items. The computation goes on in the datastore head and then the results get shipped across the wire to the web server so it can serve its request. However, Jini only supports Java leaving us no direct way to connect to the datastore using Python.

After looking around for a bit, I stumbled upon a project called JPype, which essentially allows you to launch a JVM inside Python. This allows you to instantiate and use Java objects in a completely transparent way from within Python. Using JPype, I built two modules which together, allow very simple access to the datastore though Python.

  • AuraBridge: A Java implementation of the Aura datastore interface. The bridge knows about the actual datastore because it can locate it and talk to it using Jini.
  • pyaura: A set of Python helper functions (mostly automatic type conversion). pyaura instantiates an AuraBridge instance using JPype and uses it as a proxy to get data to and from the datastore.

Example

To demonstrate how things become easy when using pyaura, imagine you are running an Aura datastore and have collected a lot of artist and tag information from the web. You might be interested in quickly seeing the number of artists that have generally been tagged by the each individual tag you know about. With these few lines of code, you can get a nice histogram that answers just that questions:

The above code produces the following plot:

tag_counts_histo

This is the result we expect, as this was generated with a datastore containing 100,000 artists. As less and less popular artists are added to the datastore, the effects of sparsity in social data kick in. Less popular artists are indeed tagged with less tags than popular artists, leading to the situation where very few tags were applied to more than 5000 artists.

This is a small example but it shows the simplicity of using pyaura. With very few lines of code, you can do pretty much anything with the data stored in Aura. This hopefully will make the Aura datastore more accessible and attractive to projects looking to take advantage of both its scalability and raw power as well as have the flexibility to quickly hack on top of it.

pfSense : a software alternative to your old router/firewall

My old D-Link router, like pretty much every other router I’ve ever owned, wasn’t very reliable in some way and so I was looking for open-source alternative firmwares like Tomato to flash it with. With the clear lack of effort put into the official firmwares, I thought it couldn’t hurt to try. Unfortunately, my router wasn’t supported by any third party firmware.

During my search, I however stumbled upon pfSense, a Free-BSD based router/firewall distro. It’s small (<100mb), runs on a 100MHz PC and includes all the features you would get on a very expensive commercial router (Firewall, NAT, VPN server, usage graphs, dynamic DNS support, per-ip bandwidth usage, QoS, etc).

Throughput on WAN interface

I already had a dedicated fileserver so I installed pfSense as a VM on it using VMWare (I could also have done it with VirtualBox, a free alternative to VMWare). All you need are two NICs. I now only use my old router as a wireless access point because pfSense naturally has a DHCP server. I could even completely let go of my D-Link router if I added a wireless NIC in my server.

If you have an old PC lying around or one that could be a host to a pfSense VM, all you might need is an extra NIC to get an enterprise-grade router that will cooperate a lot more than any cheap 50$ D-Link/Linksys/Netgear/etc router.

How does your web page look in every browser? (Updated)

Every web designer knows that making a web page come out just right in every browser can cause quite a headache, especially when combining elements like W3C standards and IE6. It’s hard to have a working copy of all the different browsers and all the different versions to test. Browsershots.org to the rescue!

Browsershots makes screenshots of your web design in different browsers. It is a free open-source online service created by Johann C. Rocholl. When you submit your web address, it will be added to the job queue. A number of distributed computers will open your website in their browser. Then they will make screenshots and upload them to the central server here.

Works great and they have 15 different browsers running on Linux, Mac OS, Windows and FreeBSD.

My blog came out perfectly on most browsers and platforms, even exotic ones like Kazehakase. Luckily for me, Microsoft was there to save the day, or else I wouldn’t have had any pictures to show.

Update : Anoter website, IE NetRenderer, allows you to get instant screenshots of your site using different version of IE.

Useful free Mac apps

I’m a relatively new Mac user so I’m keeping a list of some useful free apps that I’m using on my Mac. It’s a work in progress…

  • Instant Messaging : Adium
  • PDF Annotation : Skim
  • Notes taking : Freemind
  • Linux package manager : Macports (lots of my friends use fink)
  • EquationService : Create equations from latex that can be used in Keynote
  • MacFUSE : MacFUSE is software that allows you to write arbitrary file systems as user-space programs
  • Creative MP3 Player support : XNJB
  • OSX Ext2 Filesystem
  • Switch : Audio file converter
  • Cyberduck : SFTP/FTP
  • Espérance DV : Create a ramdisk