I attended the European Study Groups with Industry (ESGI) workshops during my PhD, first in the University of Bath in 2018 and then in the University of Cambridge in 2019. These workshops provide a platform for academic mathematicians to collaborate with industries on real-world challenges.

In Bath, Heathrow Airport tasked us with exploring cost-saving opportunities via renovation project scheduling. In Cambridge, our collaboration with the Defense Science and Technology Laboratory (DSTL) focused on strengthening image classifiers against adversarial attacks.

In Bath, using three years of Heathrow's project data, we identified challenges like staff shortages, weather-related disruptions, and project scope changes. Heathrow's project initiation patterns seemed arbitrary within their 4-year spending cycle. We developed a model to forecast time and cost overruns and assessed three scheduling strategies.

In Cambridge, we built image classifiers using PyTorch, categorizing potential threats based on their type and familiarity. I worked on adversarial training of a convolutional network model, ensuring it could withstand image perturbations. The team also delved into techniques like image pre-processing and attack detection.

For Heathrow, our scheduling strategies indicated potential savings up to 5% of their operational budget. In Cambridge, we found that while it's possible to strengthen classifiers against minor image perturbations, they remain vulnerable to more aggressive attacks due to the complexity of image data.

These ESGI workshops highlighted the value of applying academic knowledge to practical industry problems.