Ask David Tse what his greatest failure is, and he gives a surprising answer: his PhD thesis.
Now, David is no slouch. The LIDS alumnus and Stanford University professor holds several patents, has coauthored a seminal textbook, and received the 2017 Claude E. Shannon Award from the Institute of Electrical and Electronics Engineers (IEEE) Information Theory Society-information theory's highest honor, given for 'consistent and profound contributions' to the field.
But in 1994, his PhD research under Professors John Tsitsiklis and Robert Gallager (who received the Shannon Award himself in 1983) was veering off in entirely the wrong direction. David's work was focused on ways to design the rapidly-growing internet to be efficient and reliable, given the 'bursty' nature of the transmissions: how might the network cope with data being transmitted in the fits and starts that were so different from the constant bitrate of telephone conversations? The approach he took was based on telephone network design, focused on making the network as intelligent as possible to accommodate the complex traffic demands.
However, David, along with many others at the time, had the model backwards. It wasn't the network that had to allocate resources intelligently (a hugely complicated endeavor when dealing with intermittent bursts of data). Rather, the receiving ends had to intelligently accommodate to the congestion in the network. Today, the internet has been designed and grown based on this approach.
"It turned out to have minimal impact on how the internet was designed," David says of his work. The experience came with a practical lesson, however. David learned that addressing new challenges sometimes requires completely new perspectives and ways of thinking. "If you try to build a new thing, people often think, 'Ok, I've built this [other thing] before, can we use the same idea to adapt it to the new thing?' Often, it doesn't work."
That's when he realized he wanted his work to have a wider reach. "After graduation, I stopped doing this work [on intermittent data transmission] very quickly. The math was interesting, but the impact was limited," he says. "Being an engineer, I feel the urge to solve problems elegantly on one hand, but also convert that elegant solution to something in practice." To do this, he turned to studying how wireless channels can be used most efficiently, as their capacity is much more limited than that of a network built on wires.
David, who is originally from Hong Kong, first arrived at MIT in 1989 after finishing his undergraduate years at the University of Waterloo in Canada. It left an impression. At that time LIDS was housed in MIT's Building 35. "[The building] looked like a mental institution-seemingly a pretty grim place to spend five years. But the intellectual environment turned out to be amazing." That environment included the open, friendly atmosphere still tangible at LIDS today. Since much of the work done at LIDS is theoretical, David says, "Most of us were not in the 'lab.' Rather, we spent time together talking and brainstorming." Meanwhile, professors gave their time generously to students who had questions or wanted to talk over a problem.
Today, the young academic who wanted to make a broader impact has left his mark. David's work is now used in all modern cellular wireless standards. In 2004 he coauthored a textbook, Fundamentals ef Wireless Communication, which has guided generations of engineers and researchers. The textbook covers the theoretical and practical aspects of wireless communications-another reflection of his LIDS background.
Currently, his work is centered on information theory and its applications, including computational biology and machine learning. And, after nearly two decades on Berkeley's Electrical Engineering and Computer Sciences faculty, in 2014 he moved a few miles south to Stanford's Department of Electrical Engineering in in search of more interdisciplinary projects.
On Stanford's campus, the electrical engineering department is across the street from the medical school. This means that by crossing the street, David can pop in to see collaborators there and discuss their latest projects, which range from finding more efficient ways to sequence genomes, to teasing out meaning from genome-wide association studies, to identifying correlations between genes and physical traits.
In some ways, the combination of data science and genetics is a match made in heaven. The human genome is a sequence of billions of nucleotides, represented as letters from an alphabet of A,G,C,T. Beyond this basic genetic information, other molecules exert an influence on how physical traits manifest, by boosting or suppressing the machinery that reads our genes. And there's meaningful variation in genetic data from population to population, individual to individual, and even within the same individual over time.
Today, there's more of this genetic data available than ever before, as the cost of sequencing has dropped dramatically over the years. But making sense of it is akin to finding a needle in a haystack: while sequencing yields huge amounts of data, it's a challenge to find and identify the right signals. Here's where information theory plays a role, David says. "Can we develop interesting methods to find patterns and features that you can use to detect a particular disease, for instance?" The approach is potentially powerful. It makes it possible, for example, to track changes in a single individual over time-bringing our capabilities one step closer to personalized medicine.
Between all this research, spending time with his wife and five-year-old daughter, and delivering the Shannon Lecture at the IEEE International Symposium on Information Theory in Germany in June, David had a busy summer. Still, he makes as much time for his students as he can-his way of paying it forward after LIDS professors gave their time to him so generously. And the LIDS approach to thinking about problems has stayed with him all these years. "The specific project I worked on while at LIDS may have made little impact in the wider world, but the education and the skills in thinking about problems have been very valuable," he says. It's another thing he tries to pass on.
