Tuesday, April 5, 2011

IBM100: Impressions from San Jose Research 1956 – 1984

A tin roofed former cannery building housed IBM’s Research Lab on Julian Street, downtown San Jose in 1956 when I first came to work: An oven in the summer and leaking here and there during the rainy season.

Based on the success of the just announced 5MB RAMAC, we were going to build a photo memory, with a million times the RAMAC capacity! The concept and proposed implementation had little merit, but we were young, enthusiastic and a bit naïve. Today, it is difficult to realize that in 1956, most graduating engineering and science majors had no idea of what a computer was let alone the function of storage. The “system” competence, which made the RAMAC such a game changing success, had moved to the Development Lab. And it would take some time for the research organization to fully understand its role in IBM.

But then, there was serendipity. The first organic chemists came to SJ Research to address the problems posed by the photo memory project. These same chemists subsequently advanced photolithography which contributed to IBM’s leadership in integrated circuit technology. And along with physical chemists and physicists, they made major contributions to magnetic recording technology. They helped transform “paint” poured from a paper cup onto a spinning disc into the thin film technology, which is the basis for today’s storage media.

I have looked back and wondered what made San Jose Research such a success. It seems that hiring the most promising graduates from the best universities must be the answer. One of my responsibilities at IBM was the “Post-Doc Program.” This program brought promising young scientists and engineers to the lab from many countries where IBM had a major presence. To the lab, this was a welcome augmentation of junior staff; to the post docs an opportunity to work in a leading research lab with some of the best in their respective field. The most significant thing the foreign post docs took with them was the notion that there is another way of doing things -- the essence of creativity.

Maybe another contributor to the labs accomplishment was the less heralded role of outstanding management. Project R, the basis for Relational Database technology, would not have come about without the leadership of Leonard Liu, who succeeded in bringing together the efforts of an exceptionally talented group of individuals and recognized a window of opportunity.

I also recall the time when the entire storage products line was threatened by unexplained disk crashes. Every relevant resource in research and development company-wide was enlisted to address this problem. Major organizations and governments worldwide had stored their vital information on disk-based storage products. It is one thing to encounter a problem, another not to understand its cause.

The solution to the problem was not one arrived at by reverse engineering. Heinrich Hunziger, a physical chemist in the San Jose Research Lab had spent his time at IBM perfecting an analytical method of detecting and analyzing microscopic quantities of organic matter on surfaces - nothing to do with magnetic recording technology. When he learned of the disk crash problem he felt his technique might help shed some light. He was able to identify an unsuspected organic material which had contaminated the surface, speculate upon and confirm its source, as well as provide the basis for a system preventing this impurity from reaching disk surfaces. Basic research comes to the aid of the current product line!

Upon retiring in 1984, all that I had learned in IBM Research in the course of 28 years helped me to establish ICSI, the International Computer Science Institute, at UC Berkeley, still going strong. Much of that experience found it’s way into a book published by Springer in 1990, Managing Creativity in Science and Hi-Tech and into a Second Edition to appear end of 2011.

The most important lessons from my IBM Research experience is that what matters most to creative people drawn to an industrial research lab is not basic vs. applied research, independent or team effort, but to realize their full potential in doing something that is relevant, hence appreciated and rewarded.

Contribution by Ronald Kay, SJ Research 1956 – 1984 with detour to Ad Tech in
San Jose and Kingston, 1965-1967 and MIT, 1967-1968.

Monday, April 4, 2011

MRSA the super bug

A patient undergoes knee replacement surgery and is released from the hospital. Over time, a skin infection develops on the operated knee. The usual suspect is a bacterial infection called Methicillin-resistant Staphycoloccus aureus, or MRSA, which is so defiant against antiobiotics, that the only way to remove the bacteria is to reverse the knee replacement, bringing the patient back to square one.

There are a multitude of scenarios that we see in which people contract MRSA, and in densely populated cities like the San Francisco Bay Area, full on outbreaks occur (San Francisco ranked 3rd in the United States for most instances of MRSA in 2005).

We wanted to study MRSA for its prevalence but also because of its scientific characteristics. MRSA, called a "super bug," is very resistant to antibiotics and is also gram-positive, meaning there is a single cell membrane vs. dual in gram-negative bacteria. For us, that makes it easier to develop polymers that would penetrate the cell walls and destroy from the inside out. In our paper published yesterday in Nature Chemistry, we explain just how we did it.

Another important component of this breakthrough is that the nanostructures that we developed are electrically charged to be physically attracted to infected cells only - leaving the healthy cells alone. While work like this has been done by our peers, the key differentiating factor is that we have developed the first biodegradable agent. You can read more about it here.

In the science and technology group at IBM Research – Almaden, we’ve spent decades working on nanotechnology techniques that have improved semiconductors for electronic device storage, high-performance computing, solar technology and more. In applying some of the intellectual property learned from these areas, we've moved into entirely new business markets and are now partnering in new ways.


Photo captions:


Top, right: Lead IBM researcher James Hedrick in the lab at IBM Research - Almaden
Bottom, center: The antimicrobial agents developed by IBM Research and the Institute for Bioengineering and Nanomedicine attack the infected cells while protecting healthy red blood cells.

Content provided with contributions from Bob Miller, Manager, Advanced Organic Materials, IBM Research - Almaden.

Wall Street Journal:Big Blue's Tiny Bug Zapper
KQED Radio


Tuesday, March 22, 2011

IBMers visit San Jose high school

On March 14, 2011, 75 IBM engineers and scientists from San Jose and Silicon Valley went “back to school” and spent a day at Overfelt High School in San Jose to discuss the value of math and science and the importance a future career in engineering. This Celebration of Service event to give back to the community was a follow-on to a successful Black Technology Family event, held in October 2010 and sponsored by the Silicon Valley Black Employee Network. Impressed with IBM’s commitment to education and the student’s interest in technology, Overfelt High School Principal Vito Chiala asked IBM to return to the school to present to a broader student audience its message of the potential of technology to positively impact education and careers.

Throughout the day, IBMers talked to students about math, science, Deep QA system Watson, which just won the heralded Jeopardy! match, the value of a college education, and what it was like to be an engineer. At lunchtime, Laura Clayton McDonnell, Vice President, Public Sector West IMT, joined a group of 70 high achieving students enrolled in AP Calculus and AP U.S. History, to reinforce the message of making the right choice to continue their education and the habits that are necessary for success at school and work. With over 600 students in 15 classes hearing the message about education, thanks goes out to Lynn Guest, Energy Coordinator for San Jose and Silicon Valley Lab, for organizing the successful event and for all of the IBMers who gave their time to inspire our youth to go to college and pursue a career in math and science.

Friday, March 4, 2011

Rank aggregation research expert recognized by IEEE

The IEEE Computer Society has awarded its 2011 Technical Achievement Award to Ronald Fagin, Manager, Foundations of Computer Science at IBM Research - Almaden, "for pioneering contributions to the theory of rank and score aggregation."

What is "rank aggregation"?

Assume a set of "voters" and a set of "candidates." Each voter provides a list of the candidates, in order of preference. Rank aggregation is the problem of determining a consensus ordering (1st place, 2nd place, 3rd place, etc.), or sometimes simply 1st place. In case each voter provides not only an ordered list of his preferences for the candidates, but a numerical score for each candidate, then this problem is called "score aggregation". Again, the goal is to obtain a consensus ordering of the candidates.



Rank aggregation and score aggregation have many applications in numerous fields, including databases, economics, and information retrieval (where a search engine such as Google aims to provide the user with a “top k list,” say, the top 10 matches). As an example of rank aggregation, a potential home-buyer might have one ranking of a set of houses based on location, another ranking based on price, another ranking based on quality of construction, and so on. We can think of each of the houses as a candidate, and each criterion (such as location, price, and quality of construction) as a voter. Thus, there is one voter who focuses only on location, another voter who focuses only on price, and so on. In this case, rank aggregation would obtain an overall ranking of the houses that takes into account all of the criteria.



In a solely authored 1996 paper, Fagin introduced a new model for score aggregation, where there is a fixed scoring function, such as the median or the mean, and where the consensus ranking of the candidates is in order of their overall score, based on the scoring function; the goal is to obtain the top k winners. If the score of every candidate by every voter were known, it would in principle be straightforward to determine the top k winners by a simple computation. The challenge, as Fagin presented it, is to determine the top k winners while minimizing the number of database accesses. Fagin defined an algorithm (widely called "Fagin's Algorithm"), which he showed was optimal in a certain precise sense, and which greatly influenced future research on score aggregation.



In 2001, Fagin, with Amnon Lotem and Moni Naor, defined a new algorithm (the "Threshold Algorithm") for score aggregation, which he showed was optimal in an even stronger sense: not just in the worst case or the average case, but in every case! This paper won two important awards. First, it won the Best Paper Award at the prestigious ACM Symposium on Principles of Database Systems (a.k.a. PODS, which is the premier database theory conference), and has now been widely implemented. Second, this paper very recently won the PODS "Test-of Time" Award, for the paper from the 2001 PODS conference that is now judged, 10 years later, to have had the most impact.


Fagin has also explored other aspects of rank aggregation. For example, a 2003 paper by Fagin (with Ravi Kumar and D. Sivakumar) shows how to define a distance between top k lists, such as the distance between rankings by two different search engines, and is widely used in information retrieval settings.

Ronald Fagin is a member of the IBM Academy of Technology. He has won an IBM Corporate Award, eight IBM Outstanding Innovation Awards, an IBM Outstanding Technical Achievement Award, and two IBM key patent awards. He has published over 100 papers, and has co-authored a book on "Reasoning about Knowledge." He has served on more than 30 conference program committees, including serving as Program Committee Chair of four different conferences.

 He received his B.A. in mathematics from Dartmouth College, and his Ph.D. in mathematics from the University of California at Berkeley.

Accomplishments include:
  • Winner of the 2004 ACM SIGMOD Edgar F. Codd Innovations Award, a lifetime achievement award in databases, for "fundamental contributions to database theory"
  • IEEE Fellow for "contributions to finite-model theory and to relational database theory"
  • ACM Fellow for "creating the field of finite model theory and for fundamental research in relational database theory and in reasoning about knowledge"
  • AAAS Fellow for "fundamental contributions to computational complexity theory, database theory, and the theory of multi-agent systems"
  • Named Docteur Honoris Causa by the University of Paris
  • Named "Highly Cited Researcher" by the Institute for Scientific Information
  • Recipient of Best Paper awards at the 1985 International Joint Conference on Artificial Intelligence, the 2001 ACM Symposium on Principles of Database Systems, and the 2010 International Conference on Database Theory
About IEEE: With nearly 85,000 members, the IEEE Computer Society is the world's leading organization of computing professionals. Founded in 1946, it is the largest of IEEE's 38 societies. The Computer Society is dedicated to advancing the theory and application of computing and information technology.

The IEEE Computer Society Technical Achievement Award is given for outstanding and innovative contributions to the fields of computer and information science and engineering or computer technology, usually within the past 10, and not more than 15, years. Contributions must have significantly promoted technical progress in the field. The award will be presented at an awards dinner on May 25 in Albuquerque, New Mexico.

Saturday, February 19, 2011

Almaden mathematician named AAAS Fellow

IBM researcher Phokion Kolaitis was recently elected Fellow of the American Association for the Advancements of Science (AAAS) - an esteemed honor in which the organization's council elects those whose "efforts on behalf of the advancement of science or its applications are scientifically or socially distinguished." Dr. Kolaitis, recognized for his "distinguished contributions to logic in computer science and to the principles of database systems", always had a knack for math and logic. I spent some time with him to talk about this accomplishment, fueled mostly by, as he puts, "The drive for knowledge. I always had this curiosity and passion for advancing the frontier of knowledge."



And passion was a common theme in the discussion of Phokion's academic and career achievements. By high school, the Greece native had already decided he wanted to pursue mathematics. Once he entered University of Athens as an undergraduate, he felt science was what interested him, but that mathematics was the underlying basis for all things scientific. As he excelled in the university, he found himself focusing on the foundations of mathematics and, in particular, on mathematical logic. After completing his Ph.D. in Mathematics from UCLA, his interests evolved to applications of logic to computer science with particular emphasis on the connections between logic and databases. He became a professor of computer science at UC Santa Cruz and later on spent four years at Almaden as a research staff member and senior manager of the department of computer science principles and methodologies, also known as the "theory group."



"One of the reasons I wanted to come to IBM from academia was because I wanted to work on the foundations of information integration," explained Phokion, "in an area called schema mappings and data exchange." This area of study focuses on trying to develop methods and tools for restructuring data organized under one format into data organized into another. "It's a very old but recurrent problem in databases. With other colleagues at Almaden, we developed a framework for studying this problem that, in turn, facilitated the development of tools to ease the challenges." This work was recognized by IBM Research as an outstanding accomplishment, while at the same time some of the papers published on this topic have received hundreds of citations.



Outside of his career, Phokion has an uncommon hobby: collecting fountain pens. His collection features about 250 pens, mostly vintage pens from the United States, Germany, and the UK, but also modern pens from France, Italy, and Japan. His favorite and most precious pen is his father's - a Parker '51 with a rare broad nib.

Phokion keeps a close relationship with his children, both of whom are pursuing medical careers - his daughter Irini is currently a first-year pediatric resident at the Children's Memorial Hospital in Chicago after earning her MD at Georgetown University and his son Nicholas is a third year medical student at UC San Diego. Phokion is a fan of classical music and enjoys Greek poetry, citing his favorite poet as George Seferis, the winner of the Nobel 1963 Nobel Prize for Literature. Seferis captivated Phokion - as he says "Seferis' speech at the Nobel Banquet encapsulates my own feelings about the Greek language, poetry and the human condition."

Phokion will be presented the AAAS Fellow award during a recognition dinner in Washington, D.C. on February 19, 2011.

Friday, February 18, 2011

Great minds gather at Berkeley's Watson viewing event

The following post was written by guest contributor Steve Demello, CITRIS' Healthcare Director at UC Berkeley.


Some 300 enthusiastic students, faculty and researchers came together at Sutardja Dai Hall, the headquarters of the Center for Information Technology Research in the Interest of Society (CITRIS) at the University of California Berkeley, to watch and discuss the final episode of the Jeopardy! Challenge featuring IBM’s Watson platform. The event, distributed throughout the building and on the Internet with streaming video, included presentations and commentary about the event itself, Watson, and the future it portends for Natural Language Processing (NLP).

The event began with a welcome from Professor Paul Wright, Director of CITRIS, who underscored the importance of the Challenge and applauded IBM for its achievement. Jean Paul Jacob, CITRIS Senior Advisor and a career IBM executive, entertained the audience with a “pregame” presentation on the structure of the Jeopardy! game, the evolution and structure of Watson, and a summary of the first two days of competition.

Dr. Jacob was joined onstage by two distinguished commentators – Professor Shankar Sastry, Dean of the UC Berkeley College of Engineering, and Dr. Horst Simon, Deputy Director of the Lawrence Berkeley National Laboratory – on a set mimicking the game show itself, including Berkeley’s own version of Watson, dubbed “Chaplin”. Drs. Sastry and Simon commented before, at breaks during, and after the broadcast about the game, the technology, and future implications. They were joined in discussion by an engaged audience that included Professor David Patterson of the UC Berkeley Computer Science department and Michael Stewart, a former Jeopardy! champion.

The event concluded with reflections from Drs. Sastry and Simon, and discussion with the audience. Both commentators agreed that Watson represents a significant step forward in leveraging NLP, and that the capabilities demonstrated on Jeopardy! will be applied to a variety of areas, such as customer service. Both praised the “human ingenuity” at the heart of the technology, and Dr. Simon urged the group to consider this as an evolutionary step in a process that began in the 1940’s and 1950’s with early breakthroughs in computation. The commentators and audience both expressed their hope that IBM will share as much as possible with the broader research community, so that it can learn from what has been accomplished and to drive this exciting area of research forward.

Perhaps Professor Sastry best summed up the evening when he said that, as a technologist, “this was more exciting than the Super Bowl … with better ads.”

Thursday, February 17, 2011

Man vs Machine? Almaden friends and family gather for Watson's championship showdown


Watson wins!!!

Watson had the whole world engaged during its 3-day match against Jeopardy's best two contestants, Brad Rutter and Ken Jennings. The Almaden lab was host to over 200 IBMers and their friends and families on the night of the finale, and witnessed Watson battle its way to the winning spot, earning $1 million dollars for charity!

Here are some clips of the buzz before the game: