Slashdot is linking today to an article that considers the implications of the 10 year anniversary of the defeat of chess grandmaster Gary Kasparov by IBM’s Deep Blue computer. The article (here), written by philosopher Daniel Dennett, considers the possible differences, or lack of differences, between humans and machines. I’ve linked to other pieces considered by Daniel Dennett on this blog, and I consider him to be an articulate and fair judge over matters of this type. It is highly worth your time to read this piece and to think it over a bit.
You are currently browsing the archive for the Machine Learning category.
I can’t remember exactly how I came across this hilarious article, but I highly suggest checking it out (warning: it contains what some might deem slightly offensive language). It’s all about a guy named Jason who keeps being mistaken for a robot in his instant messaging conversations. It’s highly amusing. For those of you who don’t know, Alan Turing proposed a test in 1950 to gauge whether or not a computer can think. Here’s briefly (and incompletely) how the test goes:
- Recruit two humans, one to participate in the test and the other to judge it.
- Recruit a computer whose inventor claims it can think.
- Put the human judge in a room that contains only a device capable of receiving and sending text messages.
- Have the human judge type questions into this device which she would like to ask the human participant and “thinking” computer.
- One question at a time, the “thinking” computer and human participant answer these questions by writing out text answers and transmitting them back to the device in the room with the judge.
- If the judge cannot determine through the answers to these questions who is the human participant and who is the “thinking” computer, the computer wins and passes the test. In other words, since the computer tricked the judge, it can be said to think.
This is a simplistic version of the test, but it’s definitely the gist of it. In Turing’s paper he guessed that by the year 2000 a computer would have been built that was able to pass the test. He was wrong. Even now, in 2007, a computer hasn’t been built that has consistently passed this test. Interesting, huh? The full text of Turing’s paper which details the Turing test can be read here. The article is called “Computing Machinery and Intelligence”. It’s an accessible paper for anyone with these types of interests. And for more information regarding the Turing test, check out its Wikipedia article here.
Yesterday I read a paper by Tom Mitchell, the chair of the Machine Learning Department at Carnegie Mellon University, about the field of Machine Learning. This very readable introductory paper (a 7 page PDF document) can be found here. As stated by Dr. Mitchell, “Machine Learning seeks to answer the question ‘How can we build computer systems that automatically improve with experience, and what are the fundamental laws that govern all learning processes.’” The article surveys areas such as the current application successes of Machine Learning, active research questions, and ethical considerations in the discipline. If you’re at all interested in interdicsiplinary computer goodness, I highly suggest checking it out.
Part of what I love about the idea of Machine Learning is that it unavoidably collides computer science, statistics, philosophy, psychology, neuroscience, and related learning fields. Researchers in this area are actively exploring the overlap (which is sometimes messy, no doubt) of various disciplines, following the trail of questions. I plan on writing more about this later. Good stuff. Check it out.
Licensing
Recent Comments