Resource-Aware Neural Network Design with Python Tools
DOI:
https://doi.org/10.15680/IJCTECE.2022.0504003Keywords:
Resource-Aware Design, Neural Networks, Python Tools, Model Pruning, Quantization, Efficient Architectures, Energy Consumption, Computational EfficiencyAbstract
As deep learning models become increasingly complex, optimizing their efficiency—particularly in terms of computational resources and energy consumption—has become paramount. This paper explores the integration of resource-aware design principles into neural network development using Python-based tools. We examine strategies such as model pruning, quantization, and efficient architecture design, alongside profiling and optimization techniques, to create models that are both performant and resource-efficient. The goal is to provide a comprehensive framework for developing neural networks that meet the growing demand for sustainability in AI applications.
References
1. Tan, M., & Le, Q. V. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. arXiv preprint arXiv:1905.11946.arXiv+1Wikipedia+1
2. Yang, T.-J., Chen, Y.-H., & Sze, V. (2016). Designing energy-efficient convolutional neural networks using energy-aware pruning. arXiv preprint arXiv:1611.05128.arXiv+1Energy Estimation+1
3. MIT Energy Estimation Tool. (n.d.). Retrieved from https://energyestimation.mit.edu/Energy Estimation
4. Chainer. (n.d.). Retrieved from https://chainer.org/Wikipedia
5. EfficientNet. (n.d.). Retrieved from https://en.wikipedia.org/wiki/EfficientNet
