Silhouette: Toward Performance-Conscious and Transferable CPU Embeddings
This work addresses performance-conscious and transferable CPU embeddings for machine learning applications, but it appears incremental as it builds on existing embedding and transfer learning methods.
The paper tackles the problem of learning CPU embeddings from publicly-available performance datasets to enable transfer learning between datasets of different types and sizes, resulting in improved accuracy for target datasets.
Learned embeddings are widely used to obtain concise data representation and enable transfer learning between different data sets and tasks. In this paper, we present Silhouette, our approach that leverages publicly-available performance data sets to learn CPU embeddings. We show how these embeddings enable transfer learning between data sets of different types and sizes. Each of these scenarios leads to an improvement in accuracy for the target data set.