Loading...

CNNdroid: GPU-accelerated execution of trained deep convolutional neural networks on android

Latifi Oskouei, S. S ; Sharif University of Technology | 2016

835 Viewed
  1. Type of Document: Article
  2. DOI: 10.1145/2964284.2973801
  3. Publisher: Association for Computing Machinery, Inc , 2016
  4. Abstract:
  5. Many mobile applications running on smartphones and wear- able devices would potentially benefit from the accuracy and scalability of deep CNN-based machine learning algorithms. However, performance and energy consumption limitations make the execution of such computationally intensive algorithms on mobile devices prohibitive. We present a GPU- accelerated library, dubbed CNNdroid [1], for execution of trained deep CNNs on Android-based mobile devices. Empirical evaluations show that CNNdroid achieves up to 60X speedup and 130X energy saving on current mobile devices. The CNNdroid open source library is available for download at https://github.com/ENCP/CNNdroid
  6. Keywords:
  7. Deep convolutional neural network (CNN) ; Performance optimization ; Artificial intelligence ; Convolution ; Energy conservation ; Energy utilization ; Learning algorithms ; Learning systems ; Low power electronics ; Mobile devices ; Neural networks ; Open source software ; Open systems ; Optimization ; Software engineering ; Android ; Convolutional neural network ; Deep learning ; Low energy consumption ; Mobile GPU ; Performance optimizations ; RenderScript ; Android (operating system)
  8. Source: 24th ACM Multimedia Conference, MM 2016, 15 October 2016 through 19 October 2016 ; 2016 , Pages 1201-1205 ; 9781450336031 (ISBN)
  9. URL: http://dl.acm.org/citation.cfm?doid=2964284.2973801