Abstract

DNNs are widely used for complex tasks like image and signal processing, and they are in increasing demand for implementation on Internet of Things (IoT) devices. For these devices, optimizing DNN models is a necessary task. Generally, standard optimization approaches require specialists to manually fine-tune hyper-parameters to find a good trade-off between efficiency and accuracy. In this paper, we propose OptDNN, a software that employs innovative and automatic approaches to determine optimal hyper-parameters for pruning, clustering, and quantization. The models optimized by OptDNN have a smaller memory footprint, faster inference time, and a similar accuracy to the original models.

Citation

Giovannesi, L., Proietti Mattia, G., & Beraldi, R. (2024). OptDNN: Automatic deep neural networks optimizer for edge computing. Software Impacts, 100641. https://doi.org/https://doi.org/10.1016/j.simpa.2024.100641

@article{2024GiovannesiOptDNN,
  title = {OptDNN: Automatic deep neural networks optimizer for edge computing},
  author = {Giovannesi, Luca and {Proietti Mattia}, Gabriele and Beraldi, Roberto},
  year = {2024},
  journal = {Software Impacts},
  pages = {100641},
  doi = {https://doi.org/10.1016/j.simpa.2024.100641},
  issn = {2665-9638},
  url = {https://www.sciencedirect.com/science/article/pii/S2665963824000290},
  keywords = {Deep neural networks, DNN acceleration, DNN compression, Edge computing}
}