As part of the UDRC phase III, our second themed meeting on Deep Learning and Machine Learning was held on Thursday 14th November 2019, at Queen’s University Belfast. This event was intended for academic researchers, industrial partners and Dstl staff to learn about and discuss current trends in deep learning. The program consisted of a series of talks covering modern machine learning and its applications to current defence/security problems followed by a networking session. We also presented a Dstl Challenge competition and invited industry to give short pitches on the problems they face in this area.
Presentations
Welcome and introduction to the day, Neil Robertson, Queen’s University Belfast
Industrial Use Case and Lessons Learned, Chris Dickson, Thales: Listen to talk recording
Multi-label Classification: A Deep Sparse Decomposition Perspective, Mehrdad Yaghoobi, University of Edinburgh: Listen to talk recording
Robust Generative NNs, Nikolaos Dionelis, University of Edinburgh: Listen to talk recording
Deep Learning with Guarantees, Henry Gouk, University of Edinburgh: Listen to talk recording
Single Image Super-Resolution via CNN Architectures and TV-TV Minimization, Joao Mota, Heriot-Watt University: Listen to talk recording
Multi-modal Fusion, Shiyang Yan, Queen’s University Belfast: Listen to talk recording
Dstl Challenges
Automatic Target Detection for 3D Ground-Penetrating Radar Challenge Contact janet.forbes@ed.ac.uk to request data
Multiple Signal Tracking using a Simulated Passive Sonar Array in Real-time or Less: Download Data Zip file
