Mann-Whitney U Test
ConceptThe Mann-Whitney U Test is a non-parametric statistical test that does not assume a normal distribution and is reported as suitable for small sample sizes. In the provided evidence, it is used as a one-tailed test to evaluate whether differences between Vanilla AFL and Enhanced AFL fuzzing results are statistically significant.
First seen 5/26/2026
Last seen 5/29/2026
Evidence 2 chunks
Wiki v1
WIKI
Mann-Whitney U Test
Definition
The Mann-Whitney U Test is described in the provided evidence as a non-parametric test. It makes no assumption about normal distribution and therefore can be used with small sample sizes. [Definition]
NEIGHBORHOOD
No graph connections found for this entity yet. It may appear in future ingestion runs.
explore full graph →RELATIONSHIPS
1 connectionsThe paper uses the Mann-Whitney U Test to statistically analyze the difference between Vanilla AFL and Enhanced AFL results.
CITATIONS
7 sources7 citations — click to expand
[1] Definition: the Mann-Whitney U Test is non-parametric, assumes no normal distribution, and works for small sample sizes. Efficient Cross-Level Processor Verification using Coverage-guided Fuzzing
[2] Use in study: the cited study used the one-tailed Mann-Whitney U Test to compare Vanilla AFL and Enhanced AFL fuzzing results. Efficient Cross-Level Processor Verification using Coverage-guided Fuzzing
[3] Queue metric: #Queue values are test vectors that increase coverage and cause no execution mismatch; Enhanced AFL generated fewer #Queue test vectors on average. Efficient Cross-Level Processor Verification using Coverage-guided Fuzzing
[4] Queue test result: for #Queue, the study reports a 95% critical U threshold of 34, U=60, z=0, p=0.5, and concludes the improvement is not statistically significant. Efficient Cross-Level Processor Verification using Coverage-guided Fuzzing
[5] Unique-crash metric: #Unique-Crash values are unique test vectors that cause an execution mismatch, and Enhanced AFL generated more of them on average. Efficient Cross-Level Processor Verification using Coverage-guided Fuzzing
[6] Unique-crash test result: for #Unique-Crash, the study reports a 99% critical U threshold of 25, U=17, z=-2.8236, p=0.0024, and concludes Enhanced AFL is highly significantly better at detecting errors. Efficient Cross-Level Processor Verification using Coverage-guided Fuzzing
[7] Interpretation: the evidence shows the test being used to classify one apparent difference as not statistically significant and another result as highly significant. Efficient Cross-Level Processor Verification using Coverage-guided Fuzzing