Inferring Convolutional Neural Networks' Accuracies from Their Architectural Characterizations

Conference · Sun Dec 01 04:00:00 EST 2019

DOI:https://doi.org/10.1109/ICMLA.2019.00226· OSTI ID:1649450

Hoang, Duc ^[1]; Hamer, Jesse ^[2]; Perdue, Gabriel ^[3]; Young, Steven ^[4]; Miller, Jonathan ^[5]; Ghosh, Anushree ^[6]

---

1. Rhodes College, Memphis
2. University of Iowa
3. Fermi National Accelerator Laboratory (FNAL)
4. ORNL
5. Universidad Técnica Federico Santa María
6. Universidad Tecnica Federico Santa Maria, Chile

The challenge of choosing an appropriate convolutional neural network (CNN) architecture for specific applications and different data sets is still poorly understood in the literature. This is problematic, since CNNs have shown strong promise for analyzing scientific data from many domains including particle imaging detectors. In this paper, we proposed a systematic language that is useful for comparison between different CNN's architectures before training time. This helps us predict whether a network can perform better than a certain threshold accuracy before training up to 70% accuracy using simple machine learning models. Additionally, we found a coefficient of determination of 0.966 for an Ordinary Least Squares model in a regression task to predict accuracy of a large population of networks.

• Research Organization: Oak Ridge National Laboratory (ORNL), Oak Ridge, TN (United States)
• Sponsoring Organization: USDOE; USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR) (SC-21)
• DOE Contract Number: AC05-00OR22725
• OSTI ID: 1649450
• Country of Publication: United States
• Language: English