Click here to login if you're an NAE Member
Recover Your Account Information
Professor Thomas Kailath
2018 Simon Ramo Founders Award Winner
I am very grateful to the Academy for this signal honor, which I owe to the time and efforts of my nominator, Prof. José Moura, all those who supported it, and the members of the selection committee. But my greatest debt is to my stellar array of over one hundred doctoral and postdoctoral scholars, whose talents and diligence supported the contributions being recognized today. And I am delighted to note that today one more of them, Prof. Ali Sayed, was inducted into the Academy.
While I never had the honor of meeting Dr. Ramo, the first of his many books was of great value to me at a very critical time in my studies. I had the unexpected good fortune of coming to graduate school at MIT in 1957, at a time when few students from India, especially from families with limited means, could even dream of coming to the US. After completing my master’s degree in 1959, I was all set to accept a very attractive offer from Bell Labs. But I was persuaded—a nice story for another time—to stay on at MIT for my doctoral work.
One consequence of my late decision was that I had to take the PhD qualifying examinations much earlier than I had expected. In those more leisurely days, the examination comprised eight one-hour written tests. If one did reasonably well on them, one had to face a three-hour oral interview with four faculty members. And as that session progressed, they were surprised to discover, or rather uncover, the shocking fact that I did not know about Maxwell’s famous equations of electromagnetic theory. How was that possible?
In India, in those pre-IIT days, one’s major, which in my case was radio engineering, was covered only in the final year, after one had completed courses in engineering drawing, workshop, surveying, statics and dynamics, and so on. And we had only one, albeit world-famous, textbook, Radio Engineering, 4th edition, by Frederick Terman of Stanford. Terman was a very notable engineer but, at the time he wrote the book, he did not believe that Maxwell’s equations would ever be useful for practicing radio engineers. So of course I failed the examination. However, since my master’s thesis had already gained some attention, the examiners kindly decided that I (and another similarly challenged and now famous friend who had failed to explain how steel was hardened) should retake the oral examination, but only after we had successfully served as teaching assistants for the MIT undergraduate course in electromagnetics. In doing so, we found that a very useful resource was a 1944 textbook, Fields and Waves in Communication Electronics, written by Simon Ramo and John Whinnery. So thanks in part to Dr. Ramo, I successfully did manage to exit honorably from MIT, in fact as the first Indian-born recipient of a PhD in electrical engineering from MIT. You can imagine my great delight as I stand before you today to receive an award bearing Dr. Ramo’s name. I also want to gratefully note here that my time at MIT laid the foundation for all of my later accomplishments. I have the fondest memories of great teachers, research advisors, and a wonderful cohort of fellow students, during what has been called the golden age of information theory.
It probably would be wise for me to stop here, especially at this point in a long afternoon, but I am going to risk venturing a few thoughts on the role of theory in engineering. This is a topic that has always been controversial. Thomas Edison had no patience with theory—and no one can challenge his successes. But I do want to emphasize a very important and not always appreciated point: even very powerful technology can often not be effectively used without a good theoretical underpinning. This became very evident during World War II, when the skills of mathematicians and physicists had to be deployed in order to effectively use several important engineering inventions. In my own career, I changed the major focus of my research roughly every decade and, each time, my students and I found that our success was based on a careful theoretical formulation of the problem at hand.
This brings me to the current buzz about (the new) AI (artificial intelligence), ML (machine learning), CNNs (convolutional neural networks), DNNs (deep neural networks), and similar abbreviations. By taking advantage of vast computational resources to analyze and process billions of bits of labeled data, remarkable and valuable achievements have been made in fields such as speech understanding, face recognition, language translation, and machines that play games such as Go and chess. I am by no means a scholar of AI, but I would like to join in the chorus of many colleagues, including important pioneers, who are concerned about signs of “irrational exuberance.” Often, when these technologies are addressed to new problems, such as evaluating job applications or in the criminal justice system, they can lead to erroneous conclusions. And even in already well-explored fields such as computer vision, existing systems can be quite fragile—small changes in the presented data can lead to dangerous errors. I hasten to note that there is a lot of serious work going on to overcome this and other limitations. However, to get a bit technical for a moment, I believe that more theoretical work will be needed to answer questions such as determining the smallest number of layers that a deep neural network will need to achieve a desired level of performance in a particular application. It would be useful to have concepts similar to Shannon’s channel capacity in information theory, which defines an upper limit beyond which reliable communication cannot be achieved.
Please indulge me as I try to drive home the value of theory by recounting, with some poetic license, parts of a story, entitled “A Matter-of-Fact Fairy Tale,” by A.A. Milne, the creator of Winnie the Pooh. Long, long ago, in a big forest, there lived a mighty giant who had agreed to visit a friend who lived 11 miles away. The giant had just acquired the latest high-tech marvel: seven-league boots (this is an English fairy tale: a league is 3 miles, so a single step would take him 21 miles). He quickly strapped them on and strode out toward his friend’s house 11 miles away. Of course he badly overshot, and after several further vain efforts, he had to take off his magic boots and glumly trudge to his goal. The problem, as the author explained, was that these were the days before Euclid, so the giant did not know that what he had to do was to construct an isosceles triangle with a base of 11 miles and sides of 21 miles. Theory to the rescue!
Finally, I must gratefully acknowledge the many opportunities made available to me in my adoptive country, with its network of research universities like no other. Over the years, I have benefited greatly from the kindnesses of special teachers and students, research sponsors, colleagues and friends. My father instilled in me a passion for lifelong learning and from my mother I gained a sense of purpose and determination. While making time for our four children and her many interests and ventures, my late wife Sarah also indulged me enough to enable the career that I have had. And, truly finally, I would like to acknowledge my wife Anu, in the audience this afternoon, who inspires me to keep searching—for fresh adventures and new discoveries.
Thank you all.