The environmental consequence of deep learning

Borowiec, Damian; Harper, Richard; Garraghan, Peter

doi:10.1093/itnow/bwab099

Cited by 2 publications

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

Trimmer: Cost-Efficient Deep Learning Auto-tuning for Cloud Datacenters

Borowiec

Yeung

Friday

et al. 2022

2022 IEEE 15th International Conference on Cloud Computing (CLOUD)

View full text Add to dashboard Cite

Cloud datacenters capable of provisioning high performance Machine Learning-as-a-Service (MLaaS) at reduced resource cost is achieved via auto-tuning: automated tensor program optimization of Deep Learning models to minimize inference latency within a hardware device. However given the extensive heterogeneity of Deep Learning models, libraries, and hardware devices, performing auto-tuning within Cloud datacenters incurs a significant time, compute resource, and energy cost of which state-of-the-art auto-tuning is not designed to mitigate. In this paper we propose Trimmer, a high performance and cost-efficient Deep Learning auto-tuning framework for Cloud datacenters. Trimmer maximizes DL model performance and tensor program cost-efficiency by preempting tensor program implementations exhibiting poor optimization improvement; and applying an ML-based filtering method to replace expensive low performing tensor programs to provide greater likelihood of selecting low latency tensor programs. Through an empirical study exploring the cost of DL model optimization techniques, our analysis indicates that 26-43% of total energy is expended on measuring tensor program implementations that do not positively contribute towards auto-tuning. Experiment results show that Trimmer achieves high auto-tuning cost-efficiency across different DL models, and reduces auto-tuning energy use by 21.8-40.9% for Cloud clusters whilst achieving DL model latency equivalent to state-of-the-art techniques.

show abstract

Trimmer: Cost-Efficient Deep Learning Auto-tuning for Cloud Datacenters

Borowiec

Yeung

Friday

et al. 2022

2022 IEEE 15th International Conference on Cloud Computing (CLOUD)

View full text Add to dashboard Cite

show abstract

Efficient galaxy classification through pretraining

Schneider,

Stenning,

Elliott

2023

Front. Astron. Space Sci.

View full text Add to dashboard Cite

Deep learning has increasingly been applied to supervised learning tasks in astronomy, such as classifying images of galaxies based on their apparent shape (i.e., galaxy morphology classification) to gain insight regarding the evolution of galaxies. In this work, we examine the effect of pretraining on the performance of the classical AlexNet convolutional neural network (CNN) in classifying images of 14,034 galaxies from the Sloan Digital Sky Survey Data Release 4. Pretraining involves designing and training CNNs on large labeled image datasets unrelated to astronomy, which takes advantage of the vast amounts of such data available compared to the relatively small amount of labeled galaxy images. We show a statistically significant benefit of using pretraining, both in terms of improved overall classification success and reduced computational cost to achieve such performance.

show abstract

The environmental consequence of deep learning

Abstract: Deep learning and artificial intelligence are often viewed as panacea technologies — ones which can decarbonise many industries. But what is the carbon cost of these systems? Damian Borowiec, Richard R. Harper and Peter Garraghan discuss.

Cited by 2 publications

References 0 publications

Trimmer: Cost-Efficient Deep Learning Auto-tuning for Cloud Datacenters

Trimmer: Cost-Efficient Deep Learning Auto-tuning for Cloud Datacenters

Efficient galaxy classification through pretraining

Contact Info

Product

Resources

About