Proj CDeepFuzz Paper Reading: COMET: Coverage-guided Model Generation For Deep Learning Library Test

Abstract

背景：已有的方法(Muffin, Lemon, Cradle) can cover at most 34.1% layer inputs, 25.9% layer parameter values, and 15.6% layer sequences.

本文：COMET
Github: https://github.com/maybeLee/COMET
Bug Type: Crash, NaN, inconsistency between the TensorFlow library and the ONNXRuntime library
Task: fuzzing API of DL Libraries
Method:

1. designs a set of mutation operators and a coverage-based search algorithm to diversify layer inputs, layer parameter values, and layer sequences in DL models
2. model synthesis

实验：
对象：ONNXRuntime, MXNet, Keras-MXNet, TF2ONNX, ONNX2PyTorch, Keras, TensorFlow, PyTorch
Competitors: Muffin, Lemon, Cradle
效果：

1. +32 bugs, 21 confirmed, 7 fixed