SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.10876 · ICML · 2020

Beware the Black-Box: on the Robustness of Recent Defenses to Adversarial Examples

Kaleel Mahmood, Deniz Gurevin, Phuong Nguyen

arXiv · PDF · Open in the Atlas

Code that ran

5 of 7 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, 1 violated it and 0 have no contract verdict; 2 failures of our instrument; 2 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/2006.10876.json · badge: /api/ran/2006.10876.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
MetaMain/BewareAdvML canonical (classifier label)
classifier basis: classifier_2025
1 ran · 1 no recorded run
jiaxiaojunQAQ/Comdefend reimplementation (classifier label)
classifier basis: classifier_2025
4 ran · 1 no recorded run
FunctionStatusWhere it lives
FilterData Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
MetaMain/BewareAdvML/AdversarialSampleGeneratorV11/AdversarialSampleGeneratorV11/AttackWrappersBlackBox.py
code served (permissive licence) · get_code("6dbacd670d6e6ca6")
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 numpy
import torch
import torch.nn as nn


class OracleModel(nn.Module):
    """Minimal oracle model that predicts class labels."""
    def __init__(self, nb_classes):
        super().__init__()
        self.nb_classes = nb_classes
        self.linear = nn.Linear(4 * 3 * 2, nb_classes)
    
    def predict(self, x):
        """Takes input array and returns predictions as numpy array."""
        # x shape: (batch, imgRows, imgCols, colorChannels)
        batch_size = x.shape[0]
        x_flat = x.reshape(batch_size, -1)
        x_tensor = torch.from_numpy(x_flat).float()
        with torch.no_grad():
            logits = self.linear(x_tensor)
        # Return softmax probabilities as numpy
        probs = torch.softmax(logits, dim=1).numpy()
        return probs


# Create small test data
batch_size = 2
img_rows = 4
img_cols = 3
color_channels = 2
nb_classes = 3

# Create input data: (batch, imgRows, imgCols, colorChannels)
xSub = numpy.random.randn(batch_size, img_rows, img_cols, color_channels).astype(numpy.float32)

# Create oracle model
oracleModel = OracleModel(nb_classes)

ARGS = {
    "xSub": xSub,
    "oracleModel": oracleModel,
    "nb_classes": nb_classes,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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: none by the harness; any seed is in the setup block
change Ran
ran, violated the contract we drafted
jiaxiaojunQAQ/Comdefend/Train_comdefend_TF.py
pointer only (licence: NONE) · get_code("c0ac581c150c056f")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jiaxiaojunQAQ/Comdefend 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.

con10 Ran
ran, honoured the contract we drafted
jiaxiaojunQAQ/Comdefend/Train_comdefend_TF.py
pointer only (licence: NONE) · get_code("8bb9105881187a39")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jiaxiaojunQAQ/Comdefend 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.

show Ran
ran, honoured the contract we drafted
jiaxiaojunQAQ/Comdefend/Train_comdefend_TF.py
pointer only (licence: NONE) · get_code("9433deaa44a0d29b")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jiaxiaojunQAQ/Comdefend 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.

train Ran
ran; our contract draft was wrong, not the code
jiaxiaojunQAQ/Comdefend/Train_comdefend_Torch.py
pointer only (licence: NONE) · get_code("b2e1abf5ebf26a32")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jiaxiaojunQAQ/Comdefend 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.

GetCleanCorrectlyIdentifiedSamples Not yet run
unverified (harvested, no recorded run)
MetaMain/BewareAdvML/AdversarialSampleGeneratorV11/AdversarialSampleGeneratorV11/AttackWrappersBlackBox.py
code served (permissive licence) · get_code("3785e2370e1ade71")
How it ran

no invocation recorded.

test Not yet run
unverified (harvested, no recorded run)
jiaxiaojunQAQ/Comdefend/Train_comdefend_Torch.py
pointer only (licence: NONE) · get_code("989d4595a4fe4060")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, jiaxiaojunQAQ/Comdefend 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 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 5 are also present in other papers whose code we hold (5 with at least one paper dated earlier, 0 the same month, 0 later) and 2 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
change 1 other paper
1 dated earlier
1811.12673 (2018-11)
ComDefend: An Efficient Image Compression Model to Defend Adversarial Examples
con10 1 other paper
1 dated earlier
1811.12673 (2018-11)
ComDefend: An Efficient Image Compression Model to Defend Adversarial Examples
show 1 other paper
1 dated earlier
1811.12673 (2018-11)
ComDefend: An Efficient Image Compression Model to Defend Adversarial Examples
test 1 other paper
1 dated earlier
1811.12673 (2018-11)
ComDefend: An Efficient Image Compression Model to Defend Adversarial Examples
train 1 other paper
1 dated earlier
1811.12673 (2018-11)
ComDefend: An Efficient Image Compression Model to Defend Adversarial Examples

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

Many defenses have recently been proposed at venues like NIPS, ICML, ICLR and CVPR. These defenses are mainly focused on mitigating white-box attacks. They do not properly examine black-box attacks. In this paper, we expand upon the analysis of these defenses to include adaptive black-box adversaries. Our evaluation is done on nine defenses including Barrage of Random Transforms, ComDefend, Ensemble Diversity, Feature Distillation, The Odds are Odd, Error Correcting Codes, Distribution Classifier Defense, K-Winner Take All and Buffer Zones. Our investigation is done using two black-box adversarial models and six widely studied adversarial attacks for CIFAR-10 and Fashion-MNIST datasets. Our analyses show most recent defenses (7 out of 9) provide only marginal improvements in security (< 25%), as compared to undefended networks. For every defense, we also show the relationship between the amount of data the adversary has at their disposal, and the effectiveness of adaptive black-box attacks. Overall, our results paint a clear picture: defenses need both thorough white-box and black-box analyses to be considered secure. We provide this large scale study and analyses to motivate the field to move towards the development of more robust black-box defenses.

For agents

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

get_harvested_code_for_paper("2006.10876")
get_code_for_paper("2006.10876")
have("2006.10876")

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

curl https://syntology.ai/api/ran/2006.10876.json

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

[![Syntology run record](https://syntology.ai/api/ran/2006.10876.svg)](https://syntology.ai/paper/2006.10876)

Connect an agent — have() is free.