SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2302.01316 · ICML · 2023

Are Diffusion Models Vulnerable to Membership Inference Attacks?

Shiqi Wang, Kaidi Xu, Jinhao Duan, Fei Kong, Xiaoshuang Shi

arXiv · PDF · Open in the Atlas

Code that ran

11 of 23 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 2 have no contract verdict; 7 failures of our instrument; 12 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2302.01316.json · badge: /api/ran/2302.01316.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
jinhaoduan/secmi community
Papers with Code archive
11 ran · 11 no recorded run
jinhaoduan/SecMI-LDM community
Papers with Code archive
0 ran · 1 no recorded run
FunctionStatusWhere it lives
MIDataset Ran
ran on a synthesized input
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("d36d04cf36d3999f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

# Create small stand-in data for members and non-members
member_data = torch.randn(3, 5)  # 3 member samples, 5 features each
nonmember_data = torch.randn(4, 5)  # 4 non-member samples, 5 features each

member_label = torch.ones(3, 1)  # labels for members
nonmember_label = torch.zeros(4, 1)  # labels for non-members

INIT_ARGS = {
    'member_data': member_data,
    'nonmember_data': nonmember_data,
    'member_label': member_label,
    'nonmember_label': nonmember_label
}

# __getitem__ is the computation method for dataset-like classes
ARGS = {
    'item': 0  # Get the first item from the dataset
}

METHOD = '__getitem__'
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
ResNet Ran
ran on a synthesized input
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("23f4defe80b11a2d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float, shape (2, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import torch.nn.functional as F

class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, in_planes, planes, stride=1):
        super(BasicBlock, self).__init__()
        self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)
        self.shortcut = nn.Sequential()
        if stride != 1 or in_planes != self.expansion * planes:
            self.shortcut = nn.Sequential(
                nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
                nn.BatchNorm2d(self.expansion * planes)
            )

    def forward(self, x):
        out = F.relu(self.bn1(self.conv1(x)))
        out = self.bn2(self.conv2(out))
        out += self.shortcut(x)
        out = F.relu(out)
        return out

class ResNet(nn.Module):

    def __init__(self, block, num_blocks, num_channels=3, num_classes=10):
        super(ResNet, self).__init__()
        self.in_planes = 64
        self.conv1 = nn.Conv2d(num_channels, 64, kernel_size=3, stride=1, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(64)
        self.layer1 = self._make_layer(block, 64, num_blocks[0], stride=1)
        self.layer2 = self._make_layer(block, 128, num_blocks[1], stride=2)
        self.layer3 = self._make_layer(block, 256, num_blocks[2], stride=2)
        self.layer4 = self._make_layer(block, 512, num_blocks[3], stride=2)
        self.linear = nn.Linear(512 * block.expansion, num_classes)

    def _make_layer(self, block, planes, num_blocks, stride):
        strides = [stride] + [1] * (num_blocks - 1)
        layers = []
        for stride in strides:
            layers.append(block(self.in_planes, planes, stride))
            self.in_planes = planes * block.expansion
        return nn.Sequential(*layers)

    def forward(self, x, return_feat=False, return_mid_feat=False):
        out = F.relu(self.bn1(self.conv1(x)))
        out = self.layer1(out)
        out = self.layer2(out)
        out = self.layer3(out)
        mid_feat = out
        out = self.layer4(out)
        out = F.avg_pool2d(out, 4)
        feat = out.view(out.size(0), -1)
        out = self.linear(feat)
        if return_feat:
            return (out, feat)
        elif return_mid_feat:
            return (out, mid_feat)
        else:
            return out

INIT_ARGS = {
    'block': BasicBlock,
    'num_blocks': [1, 1, 1, 1],
    'num_channels': 3,
    'num_classes': 5
}

ARGS = {
    'x': torch.randn(2, 3, 32, 32),
    'return_feat': False,
    'return_mid_feat': False
}
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 5) — shape (2, 5) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
ResNet18 Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("eba347d140a27299")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned ResNet. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    "num_channels": 3,
    "num_classes": 10,
    "num_blocks": [2, 2, 2, 2],
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ResNet — a ResNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ResNet output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ddim_multistep Ran
ran; our contract draft was wrong, not the code
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("8875f269d4f25f54")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        # x shape: [batch_size, 4]
        # t shape: [batch_size]
        # Return noise prediction with same shape as x
        return self.linear(x)

class FLAGSConfig:
    def __init__(self):
        self.beta_1 = 0.0001
        self.beta_T = 0.02
        self.T = 1000

model = SimpleModel()
FLAGS = FLAGSConfig()
x = torch.randn(2, 4)
t_c = 100
target_steps = [200, 300, 400]
clip = False
device = 'cpu'
requires_grad = False

ARGS = {
    'model': model,
    'FLAGS': FLAGS,
    'x': x,
    't_c': t_c,
    'target_steps': target_steps,
    'clip': clip,
    'device': device,
    'requires_grad': requires_grad
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ddim_singlestep Ran
ran; our contract draft was wrong, not the code
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("8aae53c2358ad7f0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        return self.linear(x)

class FLAGS:
    beta_1 = 0.0001
    beta_T = 0.02
    T = 1000

model = SimpleModel()
x = torch.randn(2, 4)
t_c = 100
t_target = 200

ARGS = {
    'model': model,
    'FLAGS': FLAGS,
    'x': x,
    't_c': t_c,
    't_target': t_target,
    'requires_grad': False,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_intermediate_results Ran
ran; our contract draft was wrong, not the code
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("be9ea01d28064f30")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned dict. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        # Model takes x and t, returns epsilon prediction
        return self.linear(x)

class FLAGS:
    def __init__(self):
        self.beta_1 = 0.0001
        self.beta_T = 0.02
        self.T = 1000

model = SimpleModel()

flags = FLAGS()

# Create a simple data loader: list of tuples where first element is the batch
data_loader = [
    (torch.randn(2, 4),),
    (torch.randn(2, 4),),
]

t_sec = 100
timestep = 10

ARGS = {
    'model': model,
    'FLAGS': flags,
    'data_loader': data_loader,
    't_sec': t_sec,
    'timestep': timestep,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
naive_statistic_attack Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("7f5852ef1a8b8ff1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function expects t_results dict with keys for member/nonmember diffusions and samples
# Each should be a tensor that can be moved to device and converted to float
# The function handles both 5D tensors (num_timestep, batch, channels, height, width)
# and other shapes

# Create small tensors for testing
# 5D case: (num_timestep=2, batch=2, channels=3, height=32, width=32)
member_diffusions = torch.randn(2, 2, 3, 32, 32)
member_internal_samples = torch.randn(2, 2, 3, 32, 32)
nonmember_diffusions = torch.randn(2, 2, 3, 32, 32)
nonmember_internal_samples = torch.randn(2, 2, 3, 32, 32)

t_results = {
    'member_diffusions': member_diffusions,
    'member_internal_samples': member_internal_samples,
    'nonmember_diffusions': nonmember_diffusions,
    'nonmember_internal_samples': nonmember_internal_samples,
}

ARGS = {
    't_results': t_results,
    'metric': 'l2',
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
nn_eval Ran
ran, honoured the contract we drafted
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("2329c33cd9385b7f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Build a minimal model that takes data and outputs logits
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def forward(self, x):
        return self.linear(x)

model = SimpleModel()

# Build a minimal data loader as a list of (data, label) tuples
# Each batch: data is shape (2, 4), label is shape (2,)
data_loader = [
    (torch.randn(2, 4), torch.randint(0, 2, (2,)).float()),
    (torch.randn(2, 4), torch.randint(0, 2, (2,)).float()),
]

device = 'cpu'

ARGS = {
    'model': model,
    'data_loader': data_loader,
    'device': device,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
nn_train Ran
ran, honoured the contract we drafted
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("94c0240d193981e4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

# Define a simple model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def forward(self, x):
        return self.linear(x)

# Create model instance
model = SimpleModel()

# Create optimizer
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)

# Create data loader as list of (data, label) tuples
# batch_size=2, input_dim=4, output_dim=1
data_loader = [
    (torch.randn(2, 4), torch.randn(2, 1)),
    (torch.randn(2, 4), torch.randn(2, 1)),
]

# Define ARGS dict
ARGS = {
    'epoch': 1,
    'model': model,
    'optimizer': optimizer,
    'data_loader': data_loader,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
roc Ran
ran; our contract draft was wrong, not the code
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("a1c7d50f62f751e4")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np
from sklearn import metrics

torch.manual_seed(0)

# Create small tensors for member and nonmember scores
# member_scores: scores for members (should be lower to trigger TP when threshold is high)
# nonmember_scores: scores for non-members (should be higher to trigger TN when threshold is high)
member_scores = torch.tensor([0.1, 0.2, 0.3, 0.4], dtype=torch.float32)
nonmember_scores = torch.tensor([0.6, 0.7, 0.8, 0.9], dtype=torch.float32)

ARGS = {
    "member_scores": member_scores,
    "nonmember_scores": nonmember_scores,
    "n_points": 100
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
split_nn_datasets Ran
ran; our contract draft was wrong, not the code
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("f875a34ca4b25fe9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create small test tensors for the 4D case
# Shape: (batch, channels, height, width)
member_diffusions = torch.randn(4, 3, 32, 32)
member_internal_samples = torch.randn(4, 3, 32, 32)
nonmember_diffusions = torch.randn(4, 3, 32, 32)
nonmember_internal_samples = torch.randn(4, 3, 32, 32)

t_results = {
    'member_diffusions': member_diffusions,
    'member_internal_samples': member_internal_samples,
    'nonmember_diffusions': nonmember_diffusions,
    'nonmember_internal_samples': nonmember_internal_samples,
}

ARGS = {
    't_results': t_results,
    'train_portion': 0.1,
    'batch_size': 128,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
MIACIFAR10 Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("565b5f4760c965c4")
How it ran

no invocation recorded.

MIACIFAR100 Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("a37837f442664e59")
How it ran

no invocation recorded.

MIACelebA Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("e178434e00efeaa1")
How it ran

no invocation recorded.

MIAImageFolder Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("c1bd6f416b361a68")
How it ran

no invocation recorded.

MIASTL10 Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("c8669ab03b6eb470")
How it ran

no invocation recorded.

MIASVHN Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("8ea9b61c668976e2")
How it ran

no invocation recorded.

execute_attack Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("c5d32ff1786210fb")
How it ran

no invocation recorded.

get_reverse_denoise_results Not yet run
unverified (harvested, no recorded run)
jinhaoduan/SecMI-LDM/src/mia/secmi.py
code served (permissive licence) · get_code("4f56353a85d6692e")
How it ran

no invocation recorded.

load_member_data Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("1e2284fb660b2cbc")
How it ran

no invocation recorded.

nns_attack Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("af4ecede18d692d8")
How it ran

no invocation recorded.

print_result Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("ecf81b7416f17f20")
How it ran

no invocation recorded.

secmi_attack Not yet run
unverified (harvested, no recorded run)
jinhaoduan/secmi/mia_evals/secmia.py
code served (permissive licence) · get_code("d67ce440a0baf665")
How it ran

no invocation recorded.

Also present in other papers' code

Of 23 samples lifted from this paper's linked repositories, 23 carry a function identity and 0 do not. Of the 23 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 23 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.

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

Diffusion-based generative models have shown great potential for image synthesis, but there is a lack of research on the security and privacy risks they may pose. In this paper, we investigate the vulnerability of diffusion models to Membership Inference Attacks (MIAs), a common privacy concern. Our results indicate that existing MIAs designed for GANs or VAE are largely ineffective on diffusion models, either due to inapplicable scenarios (e.g., requiring the discriminator of GANs) or inappropriate assumptions (e.g., closer distances between synthetic samples and member samples). To address this gap, we propose Step-wise Error Comparing Membership Inference (SecMI), a query-based MIA that infers memberships by assessing the matching of forward process posterior estimation at each timestep. SecMI follows the common overfitting assumption in MIA where member samples normally have smaller estimation errors, compared with hold-out samples. We consider both the standard diffusion models, e.g., DDPM, and the text-to-image diffusion models, e.g., Latent Diffusion Models and Stable Diffusion. Experimental results demonstrate that our methods precisely infer the membership with high confidence on both of the two scenarios across multiple different datasets. Code is available at https: //github.com/jinhaoduan/SecMI.

For agents

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

get_harvested_code_for_paper("2302.01316")
get_code_for_paper("2302.01316")
have("2302.01316")

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

curl https://syntology.ai/api/ran/2302.01316.json

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

[![Syntology run record](https://syntology.ai/api/ran/2302.01316.svg)](https://syntology.ai/paper/2302.01316)

Connect an agent — have() is free.