In this case study, we hear from Shuyu Li (based in the Department of Economics), who has been utilising BEAR for tackling complex environmental economics problems.
As a PhD student in Environmental Economics at the University of Birmingham, I’ve always been fascinated by the ways in which cutting-edge technology can unlock new avenues for research. My work focuses on applying Machine Learning (ML) techniques to tackle complex environmental economics problems. Whether it’s analysing the evolution of environmental policies using ChatGPT-like models or evaluating their impacts with econometric-inspired ML methods, the combination of ML and High-Performance Computing (HPC) has become indispensable to my academic journey.
Among these tools, the University of Birmingham’s HPC system, BlueBEAR, stands out as a game-changer, which has been enriching my research, helping overcome challenges, and making previously unimaginable projects not only possible but also efficient.
The Intersection of Social Science, Machine Learning, and Big Data
For social scientists, the era of “big data” represents both an incredible opportunity and a daunting challenge. Social research now extends far beyond surveys and spreadsheets, delving into vast and diverse datasets like news articles, videos, conference transcripts, and even social media imagery. Extracting meaningful insights from such data is where ML excels, but this often comes at the cost of immense computational resources.
In my case, the challenge began with my first PhD project: a textual analysis of online news articles to identify and classify content related to environmental policies. This required using advanced ML models such as BERT and T5, which demand significant computational power. Processing large datasets of text and fine-tuning these models on my personal laptop quickly proved impractical—it was like trying to cross the ocean in a canoe.
Enter BlueBEAR, which transformed my workflow. With its generous storage capabilities and powerful GPU resources, BlueBEAR not only made my work feasible but also accelerated it significantly. What once felt overwhelming became manageable, and the time saved allowed me to focus on refining my research questions and interpreting results rather than troubleshooting hardware limitations.
Why is BlueBEAR Invaluable?
BlueBEAR is more than just a “supercomputer”; it’s a comprehensive ecosystem that supports researchers at every stage of their projects. Here are a few key aspects that I’ve found particularly beneficial:
- Reliable Data Storage : Storing large datasets securely is a common concern for researchers. The BEAR Research Data Store provides a safe and scalable environment for my files, allowing me to manage terabytes of data with confidence. Knowing that my data is backed up and easily accessible has eliminated a major source of stress.
- Access to Advanced Computing Resources : Training state-of-the-art ML models is computationally expensive. BlueBEAR’s GPU resources enable me to fine-tune models like BERT and T5 efficiently, turning what could have been weeks of processing on a standard laptop into mere hours. This speed translates directly into more iterations, better results, and, ultimately, deeper insights.
- Seamless Workflow Integration : BlueBEAR integrates effortlessly into my research pipeline, supporting popular tools and platforms commonly used in ML workflows. Whether it’s running Python scripts or leveraging containerized environments like Docker, the system is flexible enough to adapt to diverse project needs.
Learning to Maximize BlueBEAR: A Community Effort
What truly sets BlueBEAR apart is the emphasis on empowering researchers to use it effectively. The system isn’t just a resource—it’s part of a vibrant community led by the Advanced Research Computing team, who go above and beyond to ensure users succeed.
Training Courses
In Chinese, there’s an old saying: “Teaching someone to fish is better than simply giving them fish.” The Advanced Research Computing team embodies this philosophy through their extensive training programs. From introductory sessions for beginners to advanced workshops for seasoned users, these courses ensure that everyone can make the most of the HPC environment.
Accessible Documentation
For those of us who prefer self-learning or troubleshooting on our own, BlueBEAR provides clear, well-organised online documentation. I’ve found answers to countless questions simply by referring to these guides, which cover everything from job submissions to troubleshooting scripts.
Responsive Support
Perhaps what I appreciate most is the team’s responsiveness. They listen to user feedback and adapt their services accordingly, creating an environment that feels collaborative and tailored to our needs. Knowing that help is always just an email away gives me the confidence to tackle ambitious projects.
Looking Ahead: The Future of Social Science Research with BlueBEAR
As social scientists increasingly embrace computational methods, the demand for HPC systems like BlueBEAR will only grow. For me, it has already opened doors to projects I never thought possible. The ability to analyze extensive datasets with sophisticated ML models has not only enhanced my research but also expanded my vision of what’s achievable in environmental economics.
Beyond my personal experience, I believe BlueBEAR represents a vital resource for researchers across disciplines. Whether you’re studying climate change, economics, public health, or history, the system’s versatility and power can help you push the boundaries of your field.
Gratitude for BlueBEAR and Its Team
In reflecting on my journey so far, I feel immense gratitude for the University of Birmingham and the Advanced Research Computing team. They’ve provided not just a technological foundation for my research but also a sense of belonging to a forward-thinking academic community. Their dedication to innovation, accessibility, and support has made my PhD experience both enriching and rewarding.
If you’re a researcher at the University of Birmingham, I can’t recommend BlueBEAR enough. It’s more than a tool—it’s a partner in discovery, and I feel fortunate to have it as a cornerstone of my work.
Through BlueBEAR, I’ve learned that great research isn’t just about asking big questions; it’s about having the right resources to find the answers. Thank you, Advanced Research Computing team, for helping me—and countless others—turn bold ideas into impactful results.
We were so pleased to hear of how Shuyu was able to make use of what is on offer from Advanced Research Computing, particularly to hear of how she has made use of BlueBEAR HPC, storage and training – if you have any examples of how it has helped your research then do get in contact with us at bearinfo@contacts.bham.ac.uk.
We are always looking for good examples of use of High Performance Computing to nominate for HPC Wire Awards – see our recent winner for more details.