AI4DB: Quickly responding mechanism of validating knobs efficacy for GaussDB.

Supervisor: Hongzhi Wang, Professor in Faculty of Computing, HIT Exploiting meta-learning mechanism to enhance model generalization with dynamically changing workload of database, with the aim of decreasing model retraining cost and improving overall knobs evaluation performance.

Specifically, we mainly focus on utilizing gradient-based meta-learning on dataset knobs importance ranking & auto-tuning. Through considering each workload performance tests as a few-data task, we propose a meta-learning knobs importance ranking algorithm aimed at reducing dataset enormous knobs under the constraint of only few accessible attached configuration & performance samples each workload. We also design a differential tree model to leverage meta-learning paradigm for dataset performance evaluation & knobs tuning.