Timezone: »

Towards Continually Learning Application Performance Models
Ray Sinurat · Sandeep Madireddy · Anurag Daram · Haryadi Gunawi · Robert Ross
Event URL: https://openreview.net/forum?id=apAsvPTeKor »

Machine learning-based performance models are increasingly being used to build critical job scheduling and application optimization decisions. Traditionally, these models assume that data distribution does not change as more samples are collected over time. However, owing to the complexity and heterogeneity of production HPC systems, they are susceptible to hardware degradation, replacement, and/or software patches, which can lead to drift in the data distribution that can adversely affect the performance models. To this end, we develop continually learning performance models that account for the distribution drift, alleviate catastrophic forgetting, and improve generalizability. Our best model was able to retain accuracy, regardless of having to learn the new distribution of data inflicted by system changes, while demonstrating a 2X improvement in the prediction accuracy of the whole data sequence in comparison to the naive approach.

Author Information

Ray Sinurat (University of Chicago)
Ray Sinurat

First-year PhD student at the Department of Computer Science at the University of Chicago advised by Prof. Haryadi S. Gunawi and Sandeep Madireddy. Currently researching at UCARE group in collaboration with Argonne National Laboratory.

Sandeep Madireddy (Argonne National Laboratory)
Anurag Daram (University of Texas at San Antonio)

I am pursuing my doctorate in Electrical Engineering at UTSA. I am interested in and have experience working on embedded platforms, FPGAs, android application development, developing dynamic and lifelong learning algorithms inspired from the brain (humans and insects), on low power and computationally constrained platforms.

Haryadi Gunawi (University of Chicago)
Robert Ross (Northwestern University)

More from the Same Authors