Research opportunities in ocean data science

Our bioacoustics data science group (led by Wu-Jung Lee, Applied Physics Lab and Valentina Staneva, eScience Institute) is looking for enthusiastic undergrad and grad students to join in the spring and summer quarters. Our work is centered around extracting knowledge from large volumes of ocean acoustics data, which contain rich information about animals ranging from zooplankton, fish, to marine mammals. The ongoing projects focus on mining water column sonar data and span a broad spectrum from developing data-driven methods, building open source software and cloud applications, to joint analysis of acoustic observations and ocean environmental variables. Our group is highly collaborative with members with diverse backgrounds and experiences, and we are committed to provide a supportive environment for group members to grow and contribute to the oceanography and data science communities.

We have several potential project topics:

  1. Big Data Visualization

Develop a cloud-hosted visualization dashboard of big ocean sonar datasets. Gain experience with software development, managing big datasets, and product deployment for use by the scientific community. Learn more: echopype, holoviz.

Prerequisites: Python/R proficiency, experience with git & Github
Bonus: data visualization, UI design, cloud computing, parallel programming

  1. Integration of Environmental and Bioacoustic Data

Link environmental and sonar data to study the change of zooplankton and fish distribution associated with upwelling or warming events off the coasts of Washington and Oregon. Gain experience in integrating diverse data sources to answer scientific questions while learning to work with state-of-the-art data science libraries. Learn more: Sonar Decomposition, Upwelling, Software Carpentry Lessons

Prerequisites: coursework in oceanography, fishery sciences, acoustics or similar, basic statistics knowledge, programming experience in Python, R, or Matlab
Bonus: experience with git & Github, time-series analysis, research on relevant topics

  1. Optimization Algorithms for Mining Time Series

Implement optimization algorithms for detecting temporal patterns in large time series. Gain experience implementing algorithms from scratch, software development, dynamical modeling, analyzing performance. Learn more: NMF, Sonar Decomposition

Prerequisites: coursework in optimization and linear algebra, proficiency in Python, R, Matlab, or Julia
Bonus: experience with dimensionality reduction, algorithm development, time series analysis/dynamical models

  1. Deep Learning for Fish Detection

Develop deep learning pipelines for detecting fish clusters in sonar data. Gain experience developing and evaluating a deep learning solution for fisheries stock assessment. Learn more: Sandeel Classification, keras example

Prerequisites: machine learning coursework including convolutional neural networks, experience with deep learning libraries (such as Keras, PyTorch, etc.)
Bonus: tuning deep learning hyperparameters, semantic/object segmentation

To apply:

Send us (leewj@uw.edu, vms16@uw.edu, subject: ocean-sonar-data-science) a resume/CV (include GitHub account if available), unofficial transcripts, and a paragraph on why you are interested in working on any of these projects, what your qualifications are, what your current academic/career goals are, and how these opportunities fit in those. Feel free to share this with others who you think may be interested.

Wu-Jung Lee, Ph.D.
Senior Oceanographer

Applied Physics Laboratory

University of Washington

wjlee@apl.washington.edu | 206-685-3904