Search Blog & its Links

Wednesday, July 28, 2010

Student Tip: The 30-second Elevator Pitch

Why you need to think up of a 30-second verbal summary of your work, and what it should deliver.

Yes you need it need it even if you are a research student! The 30-second elevator pitch about the work you do. You will asked, in some form or another, time and again, and your answer will be used to size you up as a potential collaborator, funding recipient, and future employee. So its never too early to make one up.

What is expected out of the pitch?

At the minimum, the pitch needs to answer the following questions
  1. Your name, your research group (you may include your advisor's name), and if needed, your affiliation
  2. Your sub-field of specialization in the subject you work on. Remember that you will need to specialize or generalize (zoom in or zoom out) depending on the expertise of your audience. For example, if speaking to a professor of computer science you may want to say something like "I work in distributed routing algorithms for wireless ad-hoc networks" but while speaking to a HR recruiter you may want to limit it to wireless communication and algorithms.
  3. The key benefit of of your work vs. the state of art, or, why should anyone be excited about what you do.
Yep, thats it. Remember, people may not have more than 30 seconds so make sure to deliver your pitch in that time-frame. A good first impression can open a lot of doors. Make the elevator pitch your walking verbal resume.

Tuesday, July 13, 2010

Student tip: Traits of a Good Student Researcher

Are there specific traits in people that lead to success in a research career?

My cousin asked me about what kinds of students do research and what traits make someone successful in the 'research line'. An important question, and one that any student considering R&D would (and should) ask herself. But frankly, I haven't been able to stereotype the type of students who do very well as compared to the students who don't do well. What I have seen is that most students have a mix of the following abilities:

Innovator (You like breaking things to find out how they work)
"You once put sodium metal into a water filled petri dish and delighted in the aftermath, you made paper aeroplanes out of your language composition notebooks, you have a Linux OS partition on your home PC, your idea of a news website is, you are the in-house geek, your folks bought you Lego Technik sets for each Christmas since you were 5 and you always made something completely different than the model the brochure suggested."

Driven Hard worker (You are the human incarnation of an industrious ant)
"You are honestly hard-working, you are disciplined enough to see the sun rising occasionally, you solve each problem at the end of each text book chapter and go ask your professor or TA about the ones that you cannot solve, you love extra-credit problems, you have tried to read your professor's research paper (perhaps without understanding it), you type and print your homework assignments (including equations), you buy your subsequent semester books in Christmas break and then try to read these books over the Christmas break."

Whiz kid (you are an alien-like prodigy)
"You represented your country in the International Math Olympiad, you have a perfect GPA, you are in line to get the president's gold medal at your college commencement ceremony, you can compute the 17th root of Pi in one second, your professor uses your homework assignment submission as the homework solution sheet, you landed a Fulbright scholarship, you won the spelling bee competition, and you think about Extreme value theory whenever you hear about the stock market."

Don't really know (you are a normal John/Jane Doe)
"You don't fit into any of the above categories because you never thought about all the stuff mentioned up there, you simply don't care about who you are because your Facebook profile answers that question about you, you are always something of everything but not not really a lot of something, you haven't decided your major, you have a good GPA, a good understanding of your courses, a good desire to work, but the keyword is decent and not flamboyant."

Its important to remember that these traits are not orthogonal to each other, in fact success seems to come easier to those who can combine innovating, working hard and thinking smartly rather than just 'specializing' in one of these traits. Also important is the 'don't really know' category - because building a successful career requires networking and people skills - both of which depend on other non-research folks not binning you into one of the first 3 categories.

Thursday, July 8, 2010

Cellphone Microscope: Research & Development, with Impact

A contemporary example of how an elegant research idea can make a large impact.

I came across the story of a $10 cellphone microscope, invented by a team lead by Aydogan Ozcan, an assistant professor of electrical engineering and member of the California NanoSystems Institute at the University of California, Los Angeles. A cellphone microscope can come pretty handy in areas without diagnostic laboratories - e.g. underdeveloped nations. A quick way to do a red blood count, a bacterial analysis, a sperm count, you get the idea. Unsurprisingly, the invention has caught the attention of the Melinda and Bill gates Foundation and National Science Foundation among others (earning the research group multiple research grants).

And no, the researchers behind the work are not going to remain middle-class academics; apparently they have filed patents around their invention and intend to commercialize the invention. Way to go!

Here is a short video (courtesy UCLA) show casing the invention.

Wednesday, July 7, 2010

The Big Picture on the State of Innovation

In this round-table, organized by the Churchill Club, R&D leaders debate the "innovation drought" over the past decade, i.e., too much incremental research and aversion to big-bet blue-sky research.

Josephine M. Cheng, IBM Fellow & Vice President, IBM Almaden Research Center
Judy Estrin, CEO, JLABS
Rick Rashid, Senior Vice President, Research, Microsoft
Sue Siegel, Partner, Mohr Davidow Ventures

Michael Mandel, Chief Economist, BusinessWeek

Tuesday, July 6, 2010

Speed R&D via Rapid Development

Here is an extreme case of "startup-based" R&D. Below is an interview of Dom Sagolla, Twitter co-founder, who has undertaken a new startup venture called DollarApp. The idea is simple, develop 1 iPhone application using 1 developer in 1 month, then sell it for $1. I call this speed R&D (the key is the "&").

In a nutshell:
  1. Get (idea) --> Set (program) --> Go (deploy to app store, sell).
  2. Repeat

    What do you think? Post your comments below.

Monday, July 5, 2010

Student Tip: Managing Research Risk

Scientists are not known to be risk-takers in life and so it might come as a surprise to you, but research is risky - you stand to loose a lot of time and effort when a particular research thread turns out to be a red herring. Consider this (bad) scenario for a research student: many months or even years of work, and no research result or paper to show for it. Judging success or failure is usually black and white in research: simply put, good research or bad research is decided by the prestige of the journal or conference where the research is accepted for publication. The intermediate sweat and hard work that actually goes into the research is of lesser interest to the academic community, or indeed to anyone thinking of hiring you.

At the organizational or societal level research risk is mitigated via hedging, i.e., by investing in several research teams in a multitude of topics. For example, the NSF issues 100s of grants to 100s of research groups across the US every year. A few research grants that yield good results cover up for the other average and under- performing research projects. Can the same technique be applied at the individual level? But can you hedge against say, a Ph.D. topic, that seems promising to start with but may not yield tangible results two years down the line?

The answer to the question is yes. Individual research risk can be managed by putting your eggs in multiple baskets, i.e., by spreading your bets. This can be done, for example, by broadening the scope of the research problem you are investigating, trying to start up research collaborations with other researchers (often by bringing a particular skill or contribution to someone else's work), and most importantly, by being prepared to admit quickly that your research problem as being intractable and adapting.

Broadening the scope of the research problem should be done as and when new information is available. For example, if you discover that the particular algorithm you are developing is unsuitable for its primary use case, try to see if it can be adopted for another use case. You can also contribute by comparing your approach to other competing approaches in solving a research problem. For example, one approach may optimize for fast data transmission at the expense of higher battery usage in a mobile phone. Many times it is impossible to come up with a solution to a research problem that optimizes some parameters but remember that different people are looking to optimize different parameters. Doing a honest through survey of the pros-and-cons of different approaches to solve a research problem is highly relevant research, much more so than a paper describing a failed research attempt.

Perhaps the methodology you used for experimentation had some novel aspects that may be interesting to the research community. For example, in computer engineering there is a whole research community who build and discuss testbeds for doing further research. If you have been creative in your research work, then try to identify the innovative aspect and present this aspect to the relevant community.

More can be done when you start working with others. Look around in your organization or department to find folks who may have similar research interests as you do. The idea is reach outside of your comfort zone (your academic adviser) and collaborate with other people who may be working on related problems. While you may not lead any other research threads you find, working on different problems with different people is a significant value addition to your early research career. If you participate in multiple research threads, chances are that at least a few will yield great results. You have just hedged your research risk. Off course, one must be mindful of not committing ones time on too many research threads, be mindful of keeping ones academic adviser (i.e. stipend source :-) ) informed, and mindful of prioritizing which research thread is more important than the other.

Saturday, July 3, 2010

Notable Read: Technology and Courage by Ivan Sutherland

I read the essay titled "Technology and Courage" by Ivan Sutherland recently. Its one of those essays that you wished you'd read earlier. This one offers plenty of perspective into how to take on research risks, have the courage to stand by your ideas, and more importantly, the courage to stop working on them when the need arises.

Ivan Sutherland, winner of the ACM Turing award, is a noted entrepreneur and researcher in the field of computer graphics. I highlight a few interesting points from his essay here. I'd have to write down the whole essay here before I can do it justice and so I highly recommend you read it on your own (again, you can download it here).
  • There are several ways of mitigating research risk and entrepreneurial risk. One effective ways is writing things down e.g. a thesis outline or a business plan. This brings up some concrete future plan - one that you may not follow - but something to bolster courage and start working toward a goal
  • Being ignorant of the risks is good, at least in the beginning, because then you try to solve problems deemed "hard" or "unsolvable" - this is where young researchers have an advantage over their experienced counterparts
  • If you aren't failing sometimes you are not doing challenging things
  • Teamwork is important in research as it is in starting up new businesses
  • Hiding an idea by not getting it published because you are unsure of its merit is foolish because you'll never be sure if you did the right thing; science and technology has progressed at this remarkable pace in recent times only because people risked their reputations by working on ideas deemed "risky" at the time
  • Overcome the fear of starting a new research problem or a new start-up company by breaking down a big problem into many smaller sub-problems.
  • Doing familiar, low-risk things instead of trying to do genuine new stuff - like starting a company or new thread of research - may seem easier but remember the opportunity cost of doing something inane instead of investing your time and resources in the new stuff that may have a much higher payback

Thursday, July 1, 2010

Student Tip: Obtaining (the Right Version of) a Research Paper

In this post I explain how and where to look for research papers.

Research papers are usually published in a journal or the proceedings of a conference or workshop. A 'proceedings' is the compendium of papers presented to the audience during the conference or workshop. If an Internet search does not yield what you are looking for, libraries should be your preferred method of obtaining research papers. You may be surprised to learn that there is often more than one version of the same research paper available online. Getting the correct version may make the subsequent task of reading the material much easier for you.

Most libraries that subscribe to scientific publications tend to archive research journals and proceedings. Moreover, many libraries have collaborations on the local, state and even national level with other libraries that allows them to pull papers from each other. Your cost is usually limited to the copy machine charges since copying research papers for research/teaching purposes clearly falls within "academic fair use". I am often surprised by the resourcefulness of librarians when it comes to locating a paper (Google doesn't even come close!). It is also helpful to remember that while a huge swathe of research papers has been scanned and stored online, scanning produces images of pages, not machine-readable text. That means search engines cannot link your query to these papers easily.

Many recent research papers are available online nowadays but a successful search may not be enough; you may be asked to pay before you can download some papers. If you are affiliated to a university or R&D organization or have access to a good library, then it is possible that the publisher of the paper has an agreement with your organization that lets you access papers for free. You can find out if this is the case by checking with the librarian or concerned HR employee.

Increasingly there is also a push to making research papers freely available via online archive services such as Arxiv and on the websites of researchers and/or their host institutions. There is no harm in trying to ascertain if the researcher's website hosts a version of the paper you are interested in. Some researchers also solicit emails from individuals requesting their papers but emailing should be your last resort. The last thing you want to do is put off a (possibly important) member of your research community. Save the email for later - perhaps for when you need help understanding the paper.

You may find multiple versions of the same paper are sometimes available. Why should this be? Because researchers submit different versions of their work to different venues. While there are rules about not resubmitting the same paper to multiple venues, newer results, more detailed descriptions, analyses, or modified experiments are all legitimate reasons to resubmit papers for publication in another venue. Sometimes this is a blessing to the reader looking for details. For example, a graduate student's 120-page thesis is probably more detailed than her research paper (her Thesis condensed into 12 pages!) On the other hand, if you are looking for the gist of a paper, search if the researcher has posted a presentation that she made based on the paper. Spend some time looking for the version that suits your purpose best.

What do you think? Post your comments below.