MSTIL: Multi-cue Shape-aware Transferable Imbalance Learning for effective graphic API recommendation

Download PDF
Open Webpage

Rong Qin, Zeyu Wang, Sheng Huang, Luwen Huangfu

Published: 2023, Last Modified: 12 Nov 2025J. Syst. Softw. 2023EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Plot2API model suffer from overfitting due to data imbalance which can be reflected in two aspects.•An optimization strategy in MSTIL is proposed to address the overfitting caused by data imbalance.•MSTIL consists of a pertraining method, a data augmentation scheme and a new loss function.•MSTIL has an average relative mAP improvement of 12.94% across the models on all datasets.