**6. Conclusions**

Existing fuzzy testing tools and methods only focus on the code coverage or the test case generation on the critical path. It is difficult to take both the code coverage and path depth detection ability into account. Therefore, this paper proposes CVDF DYNAMIC, a fuzzy testing sample generation framework based on the bi-LSTM and the genetic algorithm.

By combining the genetic algorithm and the bi-LSTM neural network, the framework has the ability of code coverage and path depth detection and has excellent comprehensive performance. This paper also proposes path depth detection ability, which is an evaluation metrics of vulnerability detection ability under critical execution path. Meanwhile, a heuristic genetic algorithm is used for simplifying the sample set. Finally, the experimental results show that CVDF DYNAMIC is feasible and effective, and its performance is improved compared with existing fuzzy testing tools, such as AFLFast and NeuFuzz in several evaluation indices. (*FPR*, *TPR*, *ACC*, Code Coverage and WDC). The reduction in the sample set further improves the efficiency of the CVDF DYNAMIC test case generation. In the future, we will further optimize the performance of CVDF DYNAMIC by optimizing the neural network structure in CVDF DYNAMIC and perfecting the iterative rules of genetic algorithm and integrate more fuzzy testing sample generation methods to further improve the code coverage and path depth detection ability.

**Author Contributions:** Conceptualization, M.M.; methodology, M.M.; software, M.M.; validation, M.M.; formal analysis, M.M.; investigation, M.M.; resources, L.H.; data curation, M.M.; writing— original draft preparation, M.M.; writing—review and editing, L.H.; visualization, M.M.; supervision, Y.Q.; project administration, L.H.; funding acquisition, L.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This paper is supported by the National Natural Science Foundation of China: 6217071437, 62072200, 62127808.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data available on request due to restrictions eg privacy or ethical. The data presented in this study are available on request from the corresponding author. The data are not publicly available due to [Data privacy issues].

**Acknowledgments:** The authors thank the anonymous reviewers for their insightful suggestions on this work.

**Conflicts of Interest:** The authors declare no conflict of interest.
