Assessing Ocean Health at a Massive Speed & Scale

The Challenge

The first-ever Data Science Bowl challenged participants to examine more than 100,000 underwater images. The data was available thanks to sophisticated image capture methods pioneered by Hatfield Marine Science Center and their research partners. They were faced with volumes of information that would have taken years to manually analyze—data collected over one day takes one year! The challenge for the data science community was developing an algorithm that would enable researchers to monitor ocean health at a speed and scale never before possible.

Predicting Ocean Health, One Plankton at a Time

Like the smallest data point that can be the key to redefining an industry, plankton are key to Earth’s massively intricate ecosystems. A large and thriving plankton population is crucial. These organisms take 25% of CO2 released from burning fossil fuels every year. They also form the foundation for marine and terrestrial food chains. Because they are susceptible to small changes in temperature or water chemistry, plankton populations serve as an indicator for broader ocean health. A drop in plankton populations can be a predictor of devastating effects on our world.

plankton

Discoveries Make an Impact

More than 1,000 teams participated in the 2014/2015 Data Science Bowl. They collectively submitted more than 15,000 solutions to the challenge. The algorithms they created are allowing rapid assessment of plankton population distributions and numbers, enabling the marine research community to monitor ocean health at an unprecedented speed and scale. These types of real-time insights have not been possible through manual identification and analysis and represent an important step forward in understanding as well as protecting the environment. The data science community benefited as well. Tutorials and sample code were used extensively for learning and skills development and insights from the competition helped advance the state of the art in computer vision and Deep Learning.

We’re excited to receive the winning algorithms from the Data Science Bowl and to test and validate these proofs of concepts in our own labs. Our hope is that we will be able to expand upon this research and, eventually, make it an open source tool for the marine research community.

—Bob Cowen, Director Hatfield Marine Science Center

Competition Results

The top prize was awarded to Team Deep Sea, a team of deep learning specialists from Ghent University. They developed the most accurate classification algorithm, beating the current state of the art by more than 10% and for representing major advances for both the marine research and data science communities.

View the Public Leaderboard for other top-ranked entries from the 2014-2015 Data Science Bowl.

Trophy Icon Second Place

Happy Lantern Festival

Second Place Winner

view

Trophy Icon First Place

≋ Deep Sea ≋

First Place Winner

view

Trophy Icon Third Place

Poisson Process

Third Place Winner

view

2015 Partners

Oregon State University Hatfield Marine Science Center

Data Science Bowl Data Provider/Partner

Bob Cowen, PhD

Bob Cowen, PhD

Professor & Director

read bio

Prior to OSU, he also held faculty positions at the University of Miami and SUNY Stonybrook. His research interests are focused on larval fish and the plankton communities upon which they depend. To better understand life on the time and space scales relevant to these organisms, he seeks novel ways to study the plankton realm. The development of the In Situ Ichtyoplankton Imaging System is not only providing unprecedented insight into life in the plankton, but allows Bob to spend untethered days away from his desk on the high seas, ‘eaves-viewing’ on the secret lives of plankters.

Cedric Guigand

Cedric Guigand

Senior Research Associate & Co-developer

read bio

Even though his background is in fish biology, he has interests in new technologies and engineering used in the fields of biological and physical oceanography. His main contribution to the In Situ Ichtyoplankton Imaging System is the development of the shadowgraph optical system. He also manages the imaging system research cruises and supervises field research activities.

Kelly Robinson, PhD

Kelly Robinson, PhD

Postdoctoral Scholar

read bio

Kelly worked at the University of S. Mississippi, earned a PhD in marine science at the University of S. Alabama, and a Master’s in fisheries and aquatic science from the University of Florida. Her research as a coastal biological oceanographer is broadly aimed at the effects of climate-driven processes on marine zooplankton production and distribution, with an emphasis on gelatinous plankton predators (i.e. jellyfish). She is focused on how climate forcing alters trophic interactions and energy transfer between marine zooplankton groups, planktivorous fish, and their predators.

Jessica Luo

Jessica Luo

PhD Student

read bio

She studied the ecology of jellyfish, larval fish, and other plankton in the ocean. She is interested in plankton community dynamics, specifically food-web interactions and the biophysical processes that cause fine-scale plankton aggregations. She hopes that this research will provide better insight into understanding how biological hot-spots form in the ocean. Prior to starting her PhD, she lived in N. California, where she was working as the ocean education coordinator at Point Reyes National Seashore. Jessica got her Bachelor’s and Master’s degrees from Stanford University in 2007, studying the chemical oceanography of Red Sea copepods.

Su Sponaugle, PhD

Su Sponaugle, PhD

Professor, Dept of Integrative Biology

read bio

Inspired to pursue marine science while growing up in Thailand, Su has retained her love of tropical oceans by studying the processes associated with population replenishment of coral reef fishes. She is particularly interested in how young fish larvae grow and survive in the plankton and then find their way back to nearshore settlement habitats to complete their life history. Su earned her MS and PhD from Stony Brook University, spending a year in between on Capitol Hill dabbling in marine policy. Prior to OSU, Su was a professor at the University of Miami and served 10 years as the editor of the international journal, Bulletin of Marine Science.

Hatfield’s External Partners

Charles Cousin

Charles Cousin

Design & Manufacturing

read bio

As an external partner for Bellamare, LLC, he is involved with the design and manufacturing of the In Situ Ichtyoplankton Imaging System vehicles, using CAD modeling, drafting, and mechanical analysis tools. He manages the manufacturing, and is responsible for its final assembly. Charles started his career designing manned submersibles, and now focuses his activities helping scientists develop custom oceanographic instruments.

National Science Foundation

National Science Foundation

Independent Federal Agency

The competition data set used was based upon work supported by the National Science Foundation under Grant No. (1419987).

NSF supports research and education across all fields of science & engineering. In fiscal year (FY) 2014, its budget was $7.2 billion. NSF reaches all 50 states, funding grants to nearly 2,000 colleges, universities and other institutions. Each year, NSF receives about 50,000 competitive requests for funding, makes about 11,500 new funding awards, and awards about $593 million in professional and service contracts.

2015 Supporting Organizations

Solving the previously impossible is not easy. You need a community to enable and empower your success. As a group we can share experiences, strategies, and information that will truly allow us to affect change at a global scale. The organizations that support the Data Science Bowl form the underpinnings of that community.

Submit Your Ideas

Help us build an unprecedented future.

We are searching for the next Data Science Bowl challenge—a problem with the potential to change the world. If selected, the power of the entire data science community will be harnessed against it.

Contact us to submit your ideas or email DataScienceBowl@bah.com. Include an overview of the problem, your contact information, a brief description of the data, and where it can be obtained.