March 29, 2021 — The Data Science Institute (DSI) sponsored LLNL’s 27th hackathon on February 11–12. Held 4 occasions a yr, these seasonal occasions deliver the computing group collectively for a 24-hour interval the place something goes: Participants can deal with particular initiatives, be taught new programming languages, develop expertise, dig into difficult duties, and extra. The winter hackathon was the DSI’s second such sponsorship. Organizers had been knowledge scientist Ryan Dana, postdoctoral researcher Sarah Mackay, and DSI administrator Jennifer Bellig. DSI director Michael Goldman opened the occasion by noting, “Hackathons are great opportunities to explore new ideas and make connections with other staff, and to both innovate and learn.”
In a brand new twist to the standard hackathon schedule, organizers supplied 4 non-obligatory displays showcasing knowledge science strategies in COVID-19 drug discovery, inertial confinement fusion, central nervous system modeling, and querying of large graphs. Participants might additionally select to attend an introductory tutorial on deep studying (DL) for picture classification. Goldman famous, “Almost every program area at the Lab has some type of data science element. The hackathon is one way to help build that community.”
Team and particular person displays on the finish of the 24-hour interval featured a variety of initiatives. Lisa Hughey, an information analytics functions developer, used the time to be taught R Shiny and construct an interactive net software. Former hackathon organizer Geoff Cleary experimented with packaging Python functions, whereas Enterprise Application Services builders Brinda Jana and Yunki Paik continued a earlier hackathon undertaking to trace radio hazardous waste materials.
Data scientists Cindy Gonzales and Luke Jaffe ran the two-hour DL tutorial, which defined find out how to carry out multi-class picture classification in Python utilizing the PyTorch library. Image classification is an issue in pc imaginative and prescient through which a mannequin acknowledges a picture and outputs a label for it. This course of can play an necessary function in quite a lot of mission-relevant situations equivalent to chemical detection, distant sensing, optics inspections, and illness prognosis.
“We designed the material so participants wouldn’t need to know anything about deep learning, machine learning in general, or computer vision,” stated Jaffe, who works in LLNL’s Global Security Computing Applications Division (GS-CAD). “We expected some level of comfort with Python, and provided links where participants could learn more about the machine learning theory we covered.”
The staff offered pattern code through Jupyter Notebook and first walked attendees by means of importing packages and organising constants and picture show utility features. Next, the tutorial explored working with photographs as arrays and tensors—i.e., how a pc “sees” a picture to be able to classify it—utilizing a CIFAR10 dataset that comprises photographs of airplanes, automobiles, birds, cats, and different automobiles and animals.
Gonzales and Jaffe went on to explain the ideas behind neural networks, logistic regression to optimize classification accuracy, and several types of gradient descent algorithms. The tutorial included step-by-step directions for utilizing PyTorch to load knowledge and create, practice, and take a look at the DL mannequin.
Both tutorial leaders are increasing their knowledge science expertise on the job. Gonzales got here to the Lab as an administrator in 2016 and later modified careers with the assistance of LLNL’s Education Assistance Program (EAP) and Data Science Immersion Program. Now a GS-CAD knowledge scientist, she is pursuing a Master’s in Data Science through a Johns Hopkins distance-learning program. Jaffe was a Lab intern who was employed full time in 2016 after incomes undergraduate and graduate levels in Computer Engineering from Northeastern University. He is now utilizing the EAP to fund PhD research in Computer Vision at UC Berkeley. The staff hopes to current their tutorial once more to the Lab’s incoming summer season interns.
Continuity Is Crucial
The Lab has held 4 hackathons nearly because the COVID-19 pandemic started, and Goldman emphasised the significance of constant the occasion. “We’ve been out of the office for almost a year. Several new staff haven’t been onsite or met colleagues in person, so virtual events are crucial,” he acknowledged.
Although on-line attendance has not been as excessive as with in-person hackathons, this winter occasion noticed a gentle participation of 30–35 hackers all through. Bellig stated, “As much as I missed the energy of an in-person hackathon, I was quite impressed with all the people who participated virtually and, once again, with the presentations and hacking accomplishments of my fellow employees.”
These circumstances haven’t dampened enthusiasm for the occasion. Dana, who joined GS-CAD in January 2020, volunteered to assist arrange the occasion although he had not attended a earlier hackathon. “I wanted to learn more about how data science is applied throughout the Lab, and network with some of incredible talent and research that is being done,” he stated. Gonzales added, “I am definitely interested in participating in future hackathons.”