An Analysis of Deep Neural Network Models for Practical Applications
This work addresses the problem of resource utilization in DNNs for practitioners, but it is incremental as it focuses on analysis rather than new methods.
The paper analyzed deep neural network models for practical applications by evaluating metrics like accuracy, memory footprint, and power consumption, finding that power consumption is independent of batch size and architecture, and accuracy and inference time have a hyperbolic relationship.
Since the emergence of Deep Neural Networks (DNNs) as a prominent technique in the field of computer vision, the ImageNet classification challenge has played a major role in advancing the state-of-the-art. While accuracy figures have steadily increased, the resource utilisation of winning models has not been properly taken into account. In this work, we present a comprehensive analysis of important metrics in practical applications: accuracy, memory footprint, parameters, operations count, inference time and power consumption. Key findings are: (1) power consumption is independent of batch size and architecture; (2) accuracy and inference time are in a hyperbolic relationship; (3) energy constraint is an upper bound on the maximum achievable accuracy and model complexity; (4) the number of operations is a reliable estimate of the inference time. We believe our analysis provides a compelling set of information that helps design and engineer efficient DNNs.