Data-Efficient Deep Learning in Theory and Practice


Artur Bekasov

Deep models based on neural networks are the driving force behind most of current developments in machine learning research and practice. An old idea fueled by modern computational resources and data quantities is demonstrating unprecedented results in computer vision, speech and natural language processing, amongst other fields.


Supervisors: Amos Storkey & Michael Gutmann


Previous: Meta-Learning approaches in Deep Neural Networks            Next: Usability & Efficiency of Probabilistic Programming Languages