What is it about?
Image classification by utilizing big neural networks is an expensive task in terms of energy consumption. in this paper, we proposed a method to use a lighter and smaller neural network whenever is possible. This decision is made based on the input image. we have shown that utilizing our method can result in significantly lower energy consumption.
Featured Image
Photo by Clint Adair on Unsplash
Why is it important?
We have shown that not all the input images need the same amount of effort for correct classification. a small and light neural network is sufficient for many inputs whereas some other inputs need more feature extractor to be classified correctly.
Perspectives
I think this article opens a way for others who also believe using the same very deep and compute-intensive neural network architecture for all the input data leads to unnecessary costs in terms of energy consumption.
mohammad ali maleki
Read the Original
This page is a summary of: An Energy-Efficient Inference Method in Convolutional Neural Networks Based on Dynamic Adjustment of the Pruning Level, ACM Transactions on Design Automation of Electronic Systems, August 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3460972.
You can read the full text:
Contributors
The following have contributed to this page
