Jade Hayes
The Defence Data Research Centre (DDRC) has announced the opening of applications for its upcoming Data Study Group (DSG)
Applications are now open for our September 16th – 20th event in Newcastle!
Applications Close: 11th August at midnight.
The DSG employs an agile sprint approach, bringing together top talent to prototype potential solutions to scientific challenges today. Participants take part in an intensive, week-long collaborative study aimed at addressing current scientific challenges. This experience not only contributes to solving real-world problems but also enhances participants’ scientific, leadership, and teamwork skills—key for those currently working in or pursuing careers in academia, government or industry.
We welcome applicants from diverse backgrounds across disciplines. industry, government, and academia. For any further questions, please contact info@ddrc.uk.
Upcoming 2024 Challenge
Machine learning (ML) has opened a vast new world of opportunity in the military domain. One area particularly impacted by this is the Intelligence, Surveillance, and Reconnaissance (ISR) domain. Whether it is training to classify terrain or detect platforms from imagery, the use of machine learning can both increase the speed of processing of information and allow it to occur more autonomously.
However, ML models, both inside and outside the ISR domain, require a significant amount of data for training and validation. Often these data may not be readily available, limited in size or diversity, or simply not feasible to collect for the domain of use. Synthetic data presents a tool to facilitate the training of ML models in this scenario of limited real data. Despite this opportunity, the contextual strengths and weaknesses of synthetic data are not yet fully understood. Understanding the impact that training ML models on synthetic data, both in terms of performance and robustness, is vital to ensure safe and reliable models are deployed.
Challenges:
The data study group will work with a large dataset of imagery taken from the air. The overall machine learning task is to detect particular objects of interest, such as vehicles. However, labelled instances of the objects are rare. In this context the data study group will focus on how synthetic data can aid the detection.
The areas of exploration for this data set are twofold:
- Challenge One: Synthetic Data Generation
- Challenge Two: Synthetic Data Validation
