mitigating the adverse impact of low data volume and class imbalance on deep learning based methods
Key Points
- DAIRNet Theme: Autonomous Processing and Reasoning
- The University of Sydney, University of Technology Sydney and Downer Defence
- Point of contact: Dr Vera Chung
This project aimed to provide a solution for mitigating the dependency of deep learning based method on large volume training data with balanced samples for all categories. The project was divided into four parts:
- Data collection and labelling;
- Transfer learning;
- One/few-shot learning; and
- Extension to similar datasets.
