Guangtao Xue

MOPAR improves efficiency and cost-effectiveness of deep learning inference on serverless platforms.

A new framework aims to reduce performance issues for latency-critical services.

This framework combines basic models with language models for better emotion recognition.