Deep learning has had a tremendous impact in computer vision, natural language processing, and audio signal processing. Due to this success, further efforts have concentrated on porting these deep models on resource-constrained devices such as phones, wearables, and, more recently, micro- controllers. While on-device inference has now been widely explored, highly fast and efficient on-device training remains an open area of investigation. In this project, you will implement and evaluate a framework to enable fast and efficient training on smartphones. Specifically, this will include some implementational works and extensive evaluation of the framework that we have extended for enabling GPU training on Android based on RenderScript.

Contact: If you are interested in this project, please contact Young D. Kwon (ydk21@cam.ac.uk), Jagmohan Chauhan (jc2161@cam.ac.uk/J.Chauhan@soton.ac.uk), and Pan Hui (panhui@ust.hk) for more details.

Reference papers (can be a good starting point)
[1] 2017. RSTensorFlow: GPU Enabled TensorFlow for Deep Learning on Commodity Android Devices. Alzantot et al.