Research map / Probabilistic and scientific modeling / Molecules and proteins
VeriFlow: Modeling Distributions for Neural Network Verification
Faried Abu Zaid, Daniel Neider, Mustafa Yalçıner
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add adjoint affine group action to coupling layers and custom radial base distribution (VeriFlow vs MACow) (ablation) | MACow · Full MNIST | NLL: -1250.627 -> -2389.460 | Improved |
| same, VeriFlow vs MACow (ablation) | MACow · CIFAR-10 | NLL: -4215.694 -> -4500.373 | Improved |
| same, VeriFlow vs MACow (ablation) | MACow · Fashion MNIST | NLL: -847.522 -> -1386.659 | Improved |
| same, VeriFlow vs MACow (ablation) | MACow · MNIST digit 1 | NLL: -1577.637 -> -3123.905 | Improved |
| verify global robustness over latent UDL once instead of per-instance local robustness | Marabou · MNIST digit 0 | cumulative verification runtime: 3.7 s vs ~170 instances | Mixed |
| verify global robustness instead of per-instance falsification | ABCrown · MNIST digit 0 | falsification runtime: 7 samples break-even | Improved |
| learned latent distribution radiality | VeriFlow · MNIST | radiality tests: p numerically 0 for uniformity test | Hurt |
| latent norm distribution calibration | VeriFlow · MNIST digit 0 | KS test: KS 0.0375 (p = 0.152) | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MACow — MaCow: Masked Convolutional Generative Flow (NeurIPS 2019) · better on 7 of 7 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Stress-Testing Neural Network Verifiers with Provably Robust Instances (NeurIPS 2026 · similarity 0.89)
- Training Verification-Friendly Neural Networks via Neuron Behavior Consistency (AAAI 2025 · similarity 0.87)
- Certified but Private: Scalable Zero-Knowledge Proofs for the Formal Verification of Neural Networks (NeurIPS 2026 · similarity 0.86)
- Clip-and-Verify: Linear Constraint-Driven Domain Clipping for Accelerating Neural Network Verification (NeurIPS 2025 · similarity 0.86)
- VNN: Verification-Friendly Neural Networks with Hard Robustness Guarantees (ICML 2024 · similarity 0.86)
- No Soundness in the Real World: On the Challenges of the Verification of Deployed Neural Networks (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
