一句话：提出忠实模型转换以提升深度学习框架互操作性，并总结其三大挑战。
本文关注深度学习框架之间的互操作性问题。为了复用已训练模型文件及训练/服务程序，需要在不同框架间进行模型转换，将源模型转换为目标框架中语义等价的表示。摘要指出忠实的模型转换是增强互操作性的有前景技术，但面临三类挑战：不同深度学习框架之间存在明显差异；由于框架方案与优化，理解源模型语义可能困难；以及框架数量众多带来大量工程开销。摘要未给出具体转换方法、系统实现或实验评估结果，主要贡献在于阐明该问题与挑战。
Deep learning (DL) has become one of the most successful machine learning techniques. To achieve the optimal development result, there are emerging requirements on the interoperability between DL frameworks that the trained model files and training/serving programs can be re-utilized. Faithful model conversion is a promising technology to enhance the framework interoperability in which a source model is transformed into the semantic equivalent in another target framework format. However, several major challenges need to be addressed. First, there are apparent discrepancies between DL frameworks. Second, understanding the semantics of a source model could be difficult due to the framework scheme and optimization. Lastly, there exist a large number of DL frameworks, bringing potential significant engineering efforts.
DOI 原文 ·
@inproceedings{paperbot5757,
title = {Enhancing the interoperability between deep learning frameworks by model conversion},
author = {Yu Liu and Cheng Chen and Ru Zhang and Tingting Qin and Xiang Ji and Haoxiang Lin and Mao Yang},
booktitle = {FSE},
year = {2020},
doi = {10.1145/3368089.3417051}
}