SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2509.20177 · NeurIPS · 2025

Generative Model Inversion Through the Lens of the Manifold Hypothesis

Bo Han, Mingyuan Zhou, Feng Liu, Tongliang Liu, Xiong Peng, Fengfei Yu

arXiv · PDF · Open in the Atlas

Code that ran

14 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 12 failures of our instrument; 2 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2509.20177.json · badge: /api/ran/2509.20177.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
tmlr-group/AlignMI canonical (classifier label)
classifier basis: census
9 ran · 0 no recorded run
sutd-visual-computing-group/Re-thinking_MI canonical (classifier label)
classifier basis: census
5 ran · 2 no recorded run
FunctionStatusWhere it lives
Pdist2 Ran
ran, honoured the contract we drafted
tmlr-group/AlignMI/high_resolution/criteria/mmd_loss.py
pointer only (licence: NONE) · get_code("0b62387de45ba23b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

completion_network_loss Ran
ran; our contract draft was wrong, not the code
sutd-visual-computing-group/Re-thinking_MI/losses.py
pointer only (licence: NONE) · get_code("8ceea87fe22a6f64")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sutd-visual-computing-group/Re-thinking_MI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

create_initial_vectors Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/attack.py
pointer only (licence: NONE) · get_code("2a8b85ddd4d1398a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

ct Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/criteria/ct_loss.py
pointer only (licence: NONE) · get_code("eca18f52e446c713")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

dconv_bn_relu Ran
ran; our contract draft was wrong, not the code
sutd-visual-computing-group/Re-thinking_MI/models/generator.py
pointer only (licence: NONE); the stored copy get_code serves, from AI-secure/GMI-Attack, is pointer only (licence: NONE) · get_code("17b094c69aebe5ed")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sutd-visual-computing-group/Re-thinking_MI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_block Ran
ran, honoured the contract we drafted
sutd-visual-computing-group/Re-thinking_MI/models/evolve.py
pointer only (licence: NONE) · get_code("ad5dbf57f3ea2633")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sutd-visual-computing-group/Re-thinking_MI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_blocks Ran
ran; our contract draft was wrong, not the code
sutd-visual-computing-group/Re-thinking_MI/models/evolve.py
pointer only (licence: NONE); the stored copy get_code serves, from VladAndronik/face_swap, is pointer only (licence: NONE) · get_code("ea7941eb1ea8ffb2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sutd-visual-computing-group/Re-thinking_MI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

guassian_kernel Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/criteria/mmd_loss.py
pointer only (licence: NONE) · get_code("d1badab353798361")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

l2_loss Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/criteria/l2_loss.py
pointer only (licence: NONE) · get_code("f0c385819d1248ac")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

l2_loss_all_pairs Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/criteria/l2_loss.py
pointer only (licence: NONE) · get_code("0dec46f9d35f2650")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

l2_norm Ran
ran; our contract draft was wrong, not the code
sutd-visual-computing-group/Re-thinking_MI/models/evolve.py
pointer only (licence: NONE) · get_code("c54fea429589425d")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sutd-visual-computing-group/Re-thinking_MI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

mmd Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/criteria/mmd_loss.py
pointer only (licence: NONE) · get_code("931e13e1f2b410bc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

pairwise_cosine_dist Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/criteria/ct_loss.py
pointer only (licence: NONE) · get_code("8c809b3027f48d74")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

poincare_loss Ran
ran; our contract draft was wrong, not the code
tmlr-group/AlignMI/high_resolution/losses/poincare.py
pointer only (licence: NONE) · get_code("9eeaa864194d2b12")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tmlr-group/AlignMI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

load_attri Not yet run
unverified (harvested, no recorded run)
sutd-visual-computing-group/Re-thinking_MI/dataloader.py
pointer only (licence: NONE) · get_code("36a55c008abe5c4b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sutd-visual-computing-group/Re-thinking_MI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

noise_loss Not yet run
unverified (harvested, no recorded run)
sutd-visual-computing-group/Re-thinking_MI/losses.py
pointer only (licence: NONE) · get_code("45a9b6bf1ce41469")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, sutd-visual-computing-group/Re-thinking_MI when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 9 are also present in other papers whose code we hold (9 with at least one paper dated earlier, 0 the same month, 0 later) and 7 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
l2_norm 29 other papers
29 dated earlier · 2 only as an assembly brick
1804.07573 (2018-04)
MobileFaceNets: Efficient CNNs for Accurate Real-Time Face Verification on Mobile Devices
1903.08527 (2019-03)
Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set
1905.11946 (2019-05)
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
and 26 more not named here
get_block 27 other papers
27 dated earlier
1804.07573 (2018-04)
MobileFaceNets: Efficient CNNs for Accurate Real-Time Face Verification on Mobile Devices
1903.08527 (2019-03)
Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set
1907.10830 (2019-07)
U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation
and 24 more not named here
get_blocks 16 other papers
16 dated earlier
1804.07573 (2018-04)
MobileFaceNets: Efficient CNNs for Accurate Real-Time Face Verification on Mobile Devices
1903.08527 (2019-03)
Accurate 3D Face Reconstruction with Weakly-Supervised Learning: From Single Image to Image Set
1907.10830 (2019-07)
U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation
and 13 more not named here
Pdist2 4 other papers
4 dated earlier · 3 only as an assembly brick
2106.07636 (2021-06)
Meta Two-Sample Testing: Learning Kernels for Testing with Limited Data
2206.08843 (2022-06)
AutoML Two-Sample Test
2402.16041 (2024-02)
Detecting Machine-Generated Texts by Multi-Population Aware Optimization for Maximum Mean Discrepancy
and 1 more not named here
completion_network_loss 4 other papers
4 dated earlier
2010.04092 (2020-10)
Knowledge-Enriched Distributional Model Inversion Attacks
2207.12263 (2022-07)
SecretGen: Privacy Recovery on Pre-trained Models via Distribution Discrimination
2402.04013 (2024-02)
Privacy Leakage on DNNs: A Survey of Model Inversion Attacks and Defenses
and 1 more not named here
dconv_bn_relu 4 other papers
4 dated earlier
2010.04092 (2020-10)
Knowledge-Enriched Distributional Model Inversion Attacks
2201.12179 (2022-01)
Plug & Play Attacks: Towards Robust and Flexible Model Inversion Attacks
2207.12263 (2022-07)
SecretGen: Privacy Recovery on Pre-trained Models via Distribution Discrimination
and 1 more not named here
load_attri 2 other papers
2 dated earlier
2010.04092 (2020-10)
Knowledge-Enriched Distributional Model Inversion Attacks
2411.08460 (2024-11)
Trap-MID: Trapdoor-based Defense against Model Inversion Attacks
noise_loss 2 other papers
2 dated earlier
2010.04092 (2020-10)
Knowledge-Enriched Distributional Model Inversion Attacks
2411.08460 (2024-11)
Trap-MID: Trapdoor-based Defense against Model Inversion Attacks
poincare_loss 1 other paper
1 dated earlier
2201.12179 (2022-01)
Plug & Play Attacks: Towards Robust and Flexible Model Inversion Attacks

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Model inversion attacks (MIAs) aim to reconstruct class-representative samples from trained models. Recent generative MIAs utilize generative adversarial networks to learn image priors that guide the inversion process, yielding reconstructions with high visual quality and strong fidelity to the private training data. To explore the reason behind their effectiveness, we begin by examining the gradients of inversion loss w.r.t. synthetic inputs, and find that these gradients are surprisingly noisy. Further analysis reveals that generative inversion implicitly denoises these gradients by projecting them onto the tangent space of the generator manifold, thereby filtering out off-manifold components while preserving informative directions aligned with the manifold. Our empirical measurements show that, in models trained with standard supervision, loss gradients often exhibit large angular deviations from the generator manifold, indicating poor alignment with class-relevant directions. This observation motivates our central hypothesis: models become more vulnerable to MIAs when their loss gradients align more closely with the generator manifold. We validate this hypothesis by designing a novel training objective that explicitly promotes such alignment. Building on this insight, we further introduce a training-free approach to enhance gradient-manifold alignment during inversion, leading to consistent improvements over state-of-the-art generative MIAs. Code will be made publicly available at https://github.com/tmlr-group/AlignMI.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2509.20177")
get_code_for_paper("2509.20177")
have("2509.20177")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2509.20177.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2509.20177.svg)](https://syntology.ai/paper/2509.20177)

Connect an agent — have() is free.