ECE Student, Professor Win Best Paper Award at NIPS Conference

by Tiffany Fox

Together with his advisor, Professor Alon Orlitsky, Ananda Theertha Suresh, a Ph.D. student in Electrical and Computer Engineering at the University of California, San Diego, won the Best Paper award at the 29th Annual Conference on Neural Information Processing Systems (NIPS). NIPS, which took place this year in Montreal, is an international gathering of researchers in the fields of machine learning and computational neuroscience.

Ananda Theertha Suresh

Suresh and Orlitsky beat out more than 1,800 submissions from researchers around the world to receive the honor for their paper titled “Competitive Distribution Estimation: Why is Good-Turing Good?” Good-Turing refers to a statistical technique developed by British computer scientist Alan Turing and his assistant I.J. Good as part of their efforts to crack German encryption codes during World War II. The technique addresses an uncertainty problem called discrete distribution estimation, and makes it possible to estimate the probability of encountering an object not previously encountered, given a set of past observations of objects with different qualities.

By way of explanation, Suresh provides a theoretical example: “Imagine you’re trying to calculate the best possible vehicle to buy, one that will move quickly over all types of terrain -- snow, desert, roads, etc. You will only be buying one vehicle, so you need to optimize for a wide distribution of probabilities.

“With the prior computational approach, known as the MinMax algorithm, you would look at a range of 100 miles and minimize the time it would take to travel on all types of terrain by different types of vehicles. In other words," he continues, "you’re trying to minimize how long it will take in a worst-case scenario. Eventually, you’d calculate that the best possible vehicle would be a tank, since a tank can travel on these worst-case scenario terrains, but a tank is not practical because it can’t travel well on the road, which is where most people are driving. Thus, with the MinMax approach, you lose functionality.”

Prof. Arlon Orlitsky

Another example is a simple coin toss. “You have a coin and you want to find out the probability of heads” explains Suresh. “In a coin toss, the worst-case distribution for heads would be close to zero, but in reality, you’re not likely to have a worst-case distribution. This is often reflected in practice, where the best MinMax optimal estimator performs not as well as some heuristics.”

In their paper, Suresh and Orlitsky showed that instead of favoring the MinMax approach, where one optimizes for the worst-case distribution, the Good-Turing frequency estimation can be used to optimize for every distribution, not the just worst-case. “So if the MinMax estimator calculates you should have a tank, the Good-Turing estimator is like having a Transformer, where whatever situation you’re in, it adapts to it.”

Applications for the Good-Turing estimator include ecology (estimating the probability of animal populations with a certain physical trait, for example) and a field known as natural language processing, which is a branch of computer science concerned with interactions between computers and human (natural) languages. The “autocorrect” function on a smartphone is one example of a probability estimator at work, where the computer tries to estimate which word the user is trying to type, given a set of past observations.

“Good and Turing came up with this estimator in World War II and people have been using it for more than 60 years,” says Orlitsky, who holds dual appointments in the departments of Electrical and Computer Engineering and Computer Science and Engineering. “This paper provides perhaps the clearest explanation to date as to why it works.”

Watch Suresh's presentation at the NIPS conference in this video:

Emerging wearable technologies presented at Center for Wearable Sensors Summit

The UC San Diego Center for Wearable Sensors (CWS) hosted its second bi-annual Wearable Sensors Summit on Thursday, Nov. 12. The event featured faculty talks, a student poster pitch competition, a poster session and networking. (View Summit agenda as a PDF.)

Partners from industry including executives from the Center for Wearable Sensor's newest member companies Dexcom and Sabic interacted with UC San Diego faculty and students. Attendees learned about some of the latest exciting research projects in the Center for Wearable Sensors ranging from skin-like electronics to smartphone-based biosensors for portable and personalized healthcare.

Faculty Presentations

One of the speakers at the event was Professor Sheng Xu, who joined the faculty in the Department of NanoEngineering this past summer. Xu spoke about his research on developing new “soft” inorganic materials for advanced electronic sensors that can be comfortably worn on the skin. Faculty talks also featured presentations from computer science and engineering professor Tajana Rosing, nanoengineering professor Darren Lipomi, and electrical engineering professor Drew Hall.

Student Posters and Research

Student researchers showed off their latest work at the Center for Wearable Sensors (CWS) Summit. During the poster pitch competition, 12 students were each given 90 seconds to pitch their research to the audience and were evaluated by judges. The first and second place awards went to Timothy O’Connor and Kirtana Rajan, respectively.

L-R: CWS Associate Director Patrick Mercier; Peter Simpson from Dexcom, a CWS member company; graduate student Timothy O'Connor, the poster-pitch competition first-place winner; Joseph Wang CWS Director; and Manish Nandi from Sabic, a CWS member company. 

L-R: CWS Associate Director Patrick Mercier; Manish Nandi from Sabic, a CWS member company; graduate student Kirtana Rajan, the poster-pitch competition second-place winner; Joseph Wang CWS Director; and Peter Simpson from Dexcom, a CWS member company. 

O’Connor pitched his research on ultra lightweight wearable solar cells, and Rajan pitched her research on electronic hand sensors that wirelessly translate American Sign Language gestures to text on a computer. Both students are members of Professor Lipomi’s research group.

Attendees had a chance to learn more about these projects and others at the afternoon student poster session.

Afternoon poster session

Jiwoong Park (electrical and computer engineering)

Christine Chan (electrical and computer engineering)

Chul Kim (bioengineering)

Alex Sun (electrical and computer engineering)

Jayoung Kim (nanoengineering)

Kirtana Rajan (nanoengineering)

Timothy O'Connor (nanoengineering)

Learn more about the Center for Wearable Sensors at UC San Diego.

Hackathon Coming to UC San Diego October 2015

Have you heard? UC San Diego will host the inaugural SDHacks competition October 2-4, 2015. We sat down with the event’s organizer, computer engineering student Ryan Hill, to get a sense of what SDHacks is all about:

Describe SD Hacks in your own words.

SD Hacks is an opportunity for undergraduates to try new things and learn by doing. It's where we encourage you to build anything and everything, with only your imagination stopping you. We've done everything we can to have top notch mentors and resources present - that leaves the students to simply work their magic. We expect everyone to leave the event with something built and something learned. And then, after the event, we hope the hackathon never really stops for them - hopefully we kindled something deeper, and our hackers keep learning and building whatever they can dream up.

Give us the details.

• Who: YOU/undergrads from all over the nation
• What: A 36-hour-long hackathon, and we'll be providing meals the whole event. Prizes will be offered by industry-leading companies who want to foster innovation in the new generation.
• Where: UC San Diego, Triton Track & Field Stadium (click here for directions)
• When: October 2-4, 2015
• Why: I wanted to bring a hackathon to UC San Diego because we are home to some of the best and brightest – why should they have to go elsewhere to use their talents?

Are there any important dates we should know about?

Registration opened June 15th, and will be closing in late August.

Anything else we should know?

So far, there are nearly 1,500 students registered. We still have room for more – and we’re still looking for sponsors to offer prizes.

More questions? Here are some more answers:

1. WHAT IS A HACKATHON? A hackathon is an event, typically lasting several days, in which a large number of people meet to engage in collaborative computer programming.
2. WHO CAN ATTEND? We welcome any and all undergraduate students from any university to attend SD Hacks.
3.  HOW BIG ARE THE TEAMS? Teams can be as large as you’d like, but only groups of up to 4 people are eligible for all prizes and team-based travel reimbursement.
4. WHAT IF I DON'T HAVE A TEAM? No problem! We'll be having a mixer right before the hack so that you can team up. No one has to hack alone.
5.  HOW MUCH DOES IT COST? The event is free to undergraduate students who make it through our registration process. Food, electricity, internet, and all other necessary equipment for your development process will be provided.
6. HOW DO I GET TO THE EVENT? Several students will be awarded direct travel reimbursements, and there will be free bussing from major locations around California.
7. WHAT IF I HAVE A QUESTION NOT ANSWERED HERE? Feel free to contact our team by sending an email to support@sdhacks.io!

SDHacks was sponsored in part by the generous support of Qualcomm Incorporated, the UC San Diego Computer Science and Engineering Department, the UC San Diego Jacobs School of Engineering, ViaSat, Perkins Coie, Cubic Corporation, iBoss Cybersecurity, Drakontas Consulting, Space and Naval Warfare Systems Center Pacific (SSC Pacific), leidos, ESET namecheap, Microsoft, IBM, Touch Tiles, Facebook, Slice, the Cyber Center of Excellence and Praxis. 

Interested in sponsoring? Send an email to sponsor@sdhacks.io. 

Cocoon Cam, the first Wellness Video Baby Monitor

Congratulations to the von Liebig Entrepreneurism Center Cocoon Cam team for graduating the NSF I-Corps Program in Washington DC!

(From left to right) Dr. Nadir Weibel (Mentor from UCSD CSE Department), Rubi Sanchez,
Pavan Kumar and Dennis Abremski (Mentor from von Liebig Entrepreneurism Center)

Recognizing that there was a need for smarter baby monitoring, team founders Pavan Kumar, Siva Nattamai and Rubi Sanchez developed Cocoon Cam to make baby parenting less worrisome through continuous tracking of heart rate, respiratory rate and temperature.

"Siva had a baby about 2 years ago," explained Pavan Kumar, one of the founders of Cocoon Cam. "He started looking for a baby monitor that could tell that his baby is okay when he is away, but was not able to find any product in the market that met his needs."

Using machine vision to measure heart and respiratory rate and infrared sensors to detect skin temperature, Cocoon Cam allows parents to view videos and receive alerts on their smartphone when their baby needs attention.

During the von Liebig I-Corps program, the team identified the market size for the baby monitor industry as well as stakeholders and channels for bringing Cocoon Cam to market. Throughout the program, the team was mentored by entrepreneur Dennis Abremski and Computer Science and Engineering research scientist and lecturer Dr. Nadir Weibel. With their help, the team was able to start building relationships with potential partners and customers. After over 100 customer interviews, the team realized that Cocoon Cam would be viable as both a consumer product and tool in the medical field. To those who want to start their own companies, the team urges startup founders to get out of the building.

"It is important for startup founders to talk to as many potential customers as possible to understand the real customer need for your product or service," Pavan explained.

Not long after completing the von Liebig I-Corps program, the team was accepted into the NSF I-Corps Program in Washington DC in April and graduated from the program in June. Since then, the Cocoon Cam team has continued working at full speed. Within the last few months, the team took home a prize at the Moxie Center's Zahn Prize Competition, launched their website, completed prototype development and began testing Cocoon Cam with customers in San Diego.

Congratulations again to this team for all their success! We know it won't be long until Cocoon Cam hits the market!

Want to know more?

See more on their website, "like" Cocoon Cam on Facebook and keep up with the Cocoon Cam team's updates on Twitter.