Browsing by Author "Johnson, Jeff"

Now showing 1 - 2 of 2
  • Thumbnail Image
    GPU-based Private Information Retrieval for On-Device Machine Learning Inference
    Lam, Maximilian; Johnson, Jeff; Xiong, Wenjie; Maeng, Kiwan; Gupta, Udit; Li, Yang; Lai, Liangzhen; Leontiadis, Ilias; Rhu, Minsoo; Lee, Hsien-Hsin S.; Reddi, Vijay Janapa; Wei, Gu-Yeon; Brooks, David; Suh, Edward (ACM, 2024-04-27)
    On-device machine learning (ML) inference can enable the use of private user data on user devices without revealing them to remote servers. However, a pure on-device solution to private ML inference is impractical for many applications that rely on embedding tables that are too large to be stored on-device. In particular, recommendation models typically use multiple embedding tables each on the order of 1-10 GBs of data, making them impractical to store on-device. To overcome this barrier, we propose the use of private information retrieval (PIR) to efficiently and privately retrieve embeddings from servers without sharing any private information. As off-the-shelf PIR algorithms are usually too computationally intensive to directly use for latency-sensitive inference tasks, we 1) propose novel GPU-based acceleration of PIR, and 2) co-design PIR with the downstream ML application to obtain further speedup. Our GPU acceleration strategy improves system throughput by more than 20× over an optimized CPU PIR implementation, and our PIR-ML co-design provides an over 5× additional throughput improvement at fixed model quality. Together, for various on-device ML applications such as recommendation and language modeling, our system on a single V100 GPU can serve up to 100, 000 queries per second—a > 100× throughput improvement over a CPU-based baseline—while maintaining model accuracy.
  • Thumbnail Image
    Technological progress in the US catfish industry
    Hegde, Shraddha; Kumar, Ganesh; Engle, Carole; Hanson, Terry; Roy, Luke A.; Cheatham, Morgan; Avery, Jimmy; Aarattuthodiyil, Suja; van Senten, Jonathan; Johnson, Jeff; Wise, David; Dahl, Sunni; Dorman, Larry; Peterman, Mark (Wiley, 2022-04)
    The US catfish industry has undergone significant technological advancements in an attempt to achieve cost efficiencies. This study monitored the progress of the adoption of alternative and complementary technologies in the US catfish industry. A 2019-2020 multi-state in-person survey in Alabama, Arkansas, and Mississippi (n = 68), revealed increased adoption of intensively aerated ponds (6,315 ha) and split ponds (1,176 ha). The adoption of alternative, more intensive, production practices has been accompanied by increased adoption of complementary technologies of fixed-paddlewheel aeration, automated oxygen monitors, and hybrid catfish. As a result, the average aeration rate in the tristate region has increased to 7.8 kW/ha with 97% of catfish farms adopting automated oxygen monitors. About 53% of the water surface area in the tristate region was used for hybrid catfish production. Fingerling producers have also adopted a feed-based, oral vaccine against Enteric Septicemia of Catfish, with 83% of the fingerling farms and 73% of the fingerling production area vaccinated against ESC in 2020. Increased adoption of productivity-enhancing technologies in the US catfish industry explains the 59% increase in foodfish productivity from 2010 to 2019. Monitoring the progress of adoption of productivity-enhancing technologies will guide researchers and Extension personnel involved in the refinement and dissemination of these technologies.