Biathlon: Harnessing Model Resilience for Accelerating ML Inference Pipelines
This addresses latency issues in user-facing ML applications, offering a practical solution for industries and data science, though it is incremental in building on known model resilience.
The paper tackles the challenge of real-time responsiveness in ML inference pipelines by introducing Biathlon, a system that leverages model resilience to approximate input features, achieving speedups of 5.3x to 16.6x with minimal accuracy loss.
Machine learning inference pipelines commonly encountered in data science and industries often require real-time responsiveness due to their user-facing nature. However, meeting this requirement becomes particularly challenging when certain input features require aggregating a large volume of data online. Recent literature on interpretable machine learning reveals that most machine learning models exhibit a notable degree of resilience to variations in input. This suggests that machine learning models can effectively accommodate approximate input features with minimal discernible impact on accuracy. In this paper, we introduce Biathlon, a novel ML serving system that leverages the inherent resilience of models and determines the optimal degree of approximation for each aggregation feature. This approach enables maximum speedup while ensuring a guaranteed bound on accuracy loss. We evaluate Biathlon on real pipelines from both industry applications and data science competitions, demonstrating its ability to meet real-time latency requirements by achieving 5.3x to 16.6x speedup with almost no accuracy loss.