Tuesday, November 15, 2011

Interdisciplinary Computing in a Big Way: The Center for Science of Information

If you followed the 3 earlier posts on An Interdisciplinary Puzzle, the Wireless Car, and finally Banana Trucks and Stock Traders, then you know that information theory was at the heart of the discussion. And if you were really curious, and followed the link hint I provided in the last of the posts, then you figured out that there is an exciting body of work in this area going on at: The Center for Science of Information. Their mission is "to advance science and technology through a new quantitative understanding of the representation, communication, and processing of information in biological, social and engineered systems". A mouthful, but a mouthful that hopefully makes sense after reading the other posts.

The Center's web pages contain an incredible amount of information about their interdisciplinary research, teaching and outreach activities. No one institution or discipline could do their work alone; through bringing experts together from across disciplines, they hope to develop new cross-cutting principles governing the storage, compression and transmission of information (examples in the earlier posts!). Nine universities participate in the Center as well as collaborators from a wide range of corporations and industries. In addition, students, undergraduate and graduate, and post-doctoral researchers have the opportunity to become immersed in this cutting edge research. 

I had a fascinating conversation about the Center with Deepak Kumar of Bryn Mawr College. Deepak is the Associate Director of Diversity and Education for the Center, as well as a professor of Computer Science. Even prior to the creation of the Center, he has been teaching interesting interdisciplinary computing courses. One example is an undergraduate course on emergence. Emergent behavior appears everywhere in the natural world and is a perfect topic for demonstrating the utility of computational modeling.

For example, here in Southern California it is amazing to watch a group of Brown Pelicans cruise along in formation just over your head and then almost as one swoop down, realign into a row and ride the wind currents of ocean waves in the surf. They never touch each other or the ocean, although they can be inches from both.

Have you ever watched a flock of birds and wondered why they never crash into one another? I wonder: How do the Pelicans all know when to turn sharply, descend in unison, line up exactly the same distance from each other, and at some point, without apparent reason, rise up together again into the sky? If I watch any one bird, I can identify how it behaves. But somehow, and this is the puzzle, a group behavior emerges. Flocking birds are a classic example of emergent behavior which can be studied through computational modeling and Deepak uses this example in his course.

If birds don't suit your fancy, there are emergent behaviors to be studied computationally in linguistics, social networks, epidemiology (just for starters). Bryn Mawr is a perfect institution to be part of the Center for Information Science team not only because of courses like this, but because they have a  minor in Computational Methods that actively collaborates with departments across the college.

I asked Deepak to tell me more about the role computer scientists play in advancing information theory. He obliged; here are a few things he shared. [I'm going somewhat technical for the rest of this paragraph] For starters, computer scientists understand algorithms and complexity. They know that how a problem is modeled will lead to various algorithms with different complexities. Which one(s) best fit the constraints and goals of the subject matter? A computer scientist can help determine what kind of processor to use, what algorithms to use, all the ins and outs of dynamic problems and dynamic programming. Computer scientists have an expertise in the classification of problems; they can identify a set of problems as of a certain type (e.g. NP-hard) and the relationship between how a problem is modeled and the resultant effect on the model. There is more, but you get the idea.

Next year Deepak will be teaching a course in the Science of Information. This class will bring the work and ideas behind the Center for Science of Information directly to his undergraduate population. The course, and related activities such as internships and mentoring, will provide opportunities for his students to interact with other institutions and personnel on the Center team. Deepak is very excited about running this course; in fact he told me that the most inspiring aspect of his role in Center activities is the opportunity to bring new superstar research to his computing students and to make it outward looking.

It is going to be very interesting to watch how all of this work develops over the next few years. 


  1. I was very caught up in this article. The value of sharing information is that it causes us to stop, think, and reflect. The example of the flock of birds was particularly interesting and caught my eye--partly because I had just discussed this example in an AI class I am teaching using Russell's and Norvig's textbook. They include the theory of emergent behavior and use the example of the flock to illustrate it; really, they only mention it is passing but it was there so I talked about it. Perhaps the interdisciplinary aspect of AI is what has always attracted me to the subject. Most of my professional career has been outside of academics. Throughout that, I always thought that I had found success in integrating technology and business goals. On reflection, the biggest successes were probably equal shares of technology, science, and business. Another recent blog I read drew reactions of elitism to the term "data scientist" which is akin to the “science of information.” My reaction to those shouting elitism was, “No.” There is science that can be applied to data and modeling is a powerful tool. Too few people understand the science and how to apply it. I am glad there are programs like The Center for Science of Information and this blog to share and celebrate the efforts of those who are pursuing a transformative idea; one that may hold the answers to many of the grand challenges facing us.

  2. Tom,

    Thanks for writing this.

    I have been thinking about your comment concerning a negative reaction elsewhere to the term "data scientist". I have been trying to envision the perspective from which this phrase could be labeled "elitist" and having a hard time seeing it. Even without knowing the context, it sounds like insecurity speaking.

    From that thought I jumped to a memory of hearing someone once say that those who do interdisciplinary work do so because they can't cut it in a single discipline. As you so succinctly put it, my reaction is "No".