So for the past 10 weeks I have been in Paris working on artificial intelligence research at Universitie Pierre et Marie Curie (UPMC). As I wrap up my work for the trip and prepare to come back to the States, I wanted to write about the research part of the trip as opposed to the previous posts I have done that have focused on spending time in the city.
I have worked for the past year with Dr. Judy Goldsmith in UK's computer science department. When I expressed interest in going abroad this summer, she arranged for me to come to Paris and work with one of her friends on research that was very similar to the on going project I had. Her friend, Dr. Patrice Perny, was also very helpful in finding out what kind of visa I would need and arranging some funding. Dr. Jinze Liu in the CS department was also very helpful with getting financial support. So with the exception of almost falling into a housing scam, getting to and settling in Paris went very smoothly.
When I was taking French classes in middle school the go-to video for when we had a substitute teacher was a documentary on the Chateau at Versailles. Seeing the same documentary five times a year did not make me want to go to Versailles anymore then I already did. Fortunately it did not dull my interest so six weeks into my stay in Paris I decided to check it out myself. Versailles (Vear-Sie) is about as close to Paris as Versailles (Ver-Sales) is to Lexington. I was accompanied on my excursion by a group of students from Chicago who were visiting Paris. The other students had museum passes that let them avoid the long line to buy tickets but since I did not I rode along with a friend of the students' teacher who drove ahead early to get her ticket.
As Versailles is very big (and incredibly takes up only a small percentage of the entire estate) it would be impossible to describe everything. The most striking feature is how ornate everything is. And I mean everything. Every wall, ceiling, fireplace, statue, whatever. If it was in the palace it looked as if an artist had spent months on it. It is easy to see why the French people rebelled against this type of extravagance using tax money.
After four weeks in Paris, I left the city for the first time to spend the weekend just outside of Geneva, Switzerland. I was going to be visiting my girlfriend's father who works for Lexmark there. As I took the train away from Paris I began to see what had become an unfamiliar sight - nature that had not been completely shaped by human hands. It's not that Paris has no grass or trees but when you see green it is usually in a park that has been manicured to perfection. The three hour train ride gave me my view of France outside of Paris. The land itself very well could have been from within the United States. It was only the interruptions of farmhouses and other man made structures that confirmed the landscapes French identity.
The day that is Memorial day in the United States is also a holiday in France. So instead of going to work today I once again set out to see more of Paris. The goal for today: L'Arc de Triomphe, Les Champs-Elysees, and Place de la Concorde. L'Arc de Triomphe was my number one must-see for my time in Paris.
So a metro-ride-across-Paris later and I was at Charles de Gaul-Etiolle metro station, walking up to street level where the arch is. As I ascended I looked to my right and the arch came into view. First impression was very much dissapointment. While very tall, the arch was not nearly as wide and I thought it looked in pictures. This gave it a misproportionate look and made it seem much less grand. As I reached ground level though I realized I was on the side of the arch and not looking at it straight on. Then I remembered from pictures that the arch is actually 4 arches supporting a platform. Together they look like one big arch. Second impression: wow!
Since Paris is my home for the next two months I am trying hard to be more Parisien than tourist. However, I am in Paris. It would be foolish to not go see the sights. So with that in mind I set out this weekend to be a tourist. First thing on Saturday I took the metro to a crepe café near the Eiffel tower. The previous weekend I had met up with friends in the UK honors program on the honors trip to Europe and we went there. Since I planned on starting my day at the Eiffel Tower I decided to return. The owner and his wife are very friendly. They speak a little English although their accents sometimes make it hard to understand. The food was delicious (as all the food has been in Paris) and an American was also in the café so I had some English conversation while I ate.
First off, fair warning this is going to be very nerdy but I find it interesting so enjoy!
In high school computer science class I was introduced to the idea of complexity. Complexity is a way to describe how efficient a program is. In other words how much computer memory does the program need and how fast can the program be completed. Complexity is described mathematically using what is called big O notation and is written as a function of time or memory in terms of the input size. For example O(n^2) means that for n number of inputs the time to complete the program increases quadratically. Big O notation is always written as a function of one term with no coefficients. Examples are O(n), O(n^3), O(log n), etc.
Obviously some of these complexities are problematic because as the size of the input increases the complexity increases much faster. For example O(n^3) with an n value of 100,000 has 10^15 operations to perform. If the computer runs through a million operations per second the program wouldn't stop running for almost 32 years! If your computer lasted for that time it would be incredibly obsolete. Clearly no one has time to run a program for this long and 100,000 isn't even that large of an input number (consider that google searches deal with billions of sites).
I've been working on a system for recommending courses to students based on the past performance of similar students. Hopefully the system will be a useful tool for students wishing to gain some idea of what courses they might do well in. However another purpose of designing this system is to compare it with an artificial intelligence advising system that is being worked on in the computer science department. The comparison would be to see how artifical intelligence does with planning for students versus a netflix style advising system (technical term is collaborative filtering).
What is amazing about this to me is the possible use of a computer as an advisor. I think of advising decisions as complex and personal. To think that artifical intelligence has come so far as to be considered for this type of process shows just how far computing has come. Think about early computers. They filled entire rooms and required intensive set up to perform very simple computations. Now we use computers in the military (automated drones), selecting movies (netflix), playing chess, winning jeapordy, and finding love (match.com, eHarmony, etc). The early computer scientist would be astounded by how much has been done in such a short time. So the question is: what is possible in the future?
Many of us are familiar with the movie rental service Netflix. One of the features of Netflix is a recommendation system that recommends movies to the user based on their past selections and the data from similar users. While the system can't always predict perfectly what you will enjoy it is fairly accurate in gauging your tastes and using them to find movies or tv shows that you would enjoy. This system uses a technique called collaborative filtering for filtering through the company's vast collection of user data. The system is similar to Amazon's recommender system although a different type of collaborative filtering.