TY - GEN
T1 - AquaSense
T2 - 21st International Symposium on Automated Technology for Verification and Analysis, ATVA 2023
AU - Zhou, Zitong
AU - Huang, Zixin
AU - Misailovic, Sasa
N1 - Acknowledgements. This research was supported in part by NSF Grants No. CCF-1846354, CCF-1956374, CCF-200888, and CCF-2217144, and C3.ai DTI research award.
PY - 2023
Y1 - 2023
N2 - We propose a novel tool, AquaSense, to automatically reason about the sensitivity analysis of probabilistic programs. In the context of probabilistic programs, sensitivity analysis investigates how the perturbation in the parameters of prior distributions affects the program’s result, i.e., the program’s posterior distribution. AquaSense leverages quantized inference, an efficient and accurate approximate inference algorithm that represents distributions of random variables with quantized intervals. AquaSense is the first tool to support sensitivity analysis of probabilistic programs that is at the same time symbolic, differentiable, and practical. Our evaluation compares AquaSense with an existing system PSense (a system that relies on fully symbolic inference). AquaSense can compute the sensitivity of all 45 parameters from 12 programs, compared to 11/45 that PSense computes. AquaSense is particularly effective on programs with continuous distributions: it achieves an average speedup of 18.10 × over PSense (which, in contrast, can solve only a handful of problems). Our evaluation shows that AquaSense computes exact results on discrete programs. On 91% of evaluated continuous parameters, AquaSense computed the sensitivity results within 40 s with high accuracy (below 5% error). The paper also discusses AquaSense’s performance-accuracy trade-offs, which can enable different operational points for programs with different input data sizes.
AB - We propose a novel tool, AquaSense, to automatically reason about the sensitivity analysis of probabilistic programs. In the context of probabilistic programs, sensitivity analysis investigates how the perturbation in the parameters of prior distributions affects the program’s result, i.e., the program’s posterior distribution. AquaSense leverages quantized inference, an efficient and accurate approximate inference algorithm that represents distributions of random variables with quantized intervals. AquaSense is the first tool to support sensitivity analysis of probabilistic programs that is at the same time symbolic, differentiable, and practical. Our evaluation compares AquaSense with an existing system PSense (a system that relies on fully symbolic inference). AquaSense can compute the sensitivity of all 45 parameters from 12 programs, compared to 11/45 that PSense computes. AquaSense is particularly effective on programs with continuous distributions: it achieves an average speedup of 18.10 × over PSense (which, in contrast, can solve only a handful of problems). Our evaluation shows that AquaSense computes exact results on discrete programs. On 91% of evaluated continuous parameters, AquaSense computed the sensitivity results within 40 s with high accuracy (below 5% error). The paper also discusses AquaSense’s performance-accuracy trade-offs, which can enable different operational points for programs with different input data sizes.
KW - Probabilistic Programming
KW - Quantized Inference
KW - Sensitivity Analysis
UR - http://www.scopus.com/inward/record.url?scp=85176010995&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85176010995&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-45332-8_16
DO - 10.1007/978-3-031-45332-8_16
M3 - Conference contribution
AN - SCOPUS:85176010995
SN - 9783031453311
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 288
EP - 301
BT - Automated Technology for Verification and Analysis - 21st International Symposium, ATVA 2023, Proceedings
A2 - André, Étienne
A2 - Sun, Jun
PB - Springer
Y2 - 24 October 2023 through 27 October 2023
ER -