SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.15147 · 2020

A General Framework For Detecting Anomalous Inputs to DNN Classifiers

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 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; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 0 failures of our instrument; 0 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/2007.15147.json · badge: /api/ran/2007.15147.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
jayaram-r/adversarial-detection canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
FunctionStatusWhere it lives
bhattacharya_coeff Ran
ran on a synthesized input
jayaram-r/adversarial-detection/expts/plot_layer_stats.py
code served (permissive licence) · get_code("2737b6a2e4410c2a")
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 float64, ndim 0, 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 numpy as np

# Create sample data for the function
np.random.seed(42)

# Generate two sets of samples from different distributions
samples1 = np.random.normal(loc=0, scale=1, size=100)
samples2 = np.random.normal(loc=2, scale=1.5, size=100)

# Create bin edges for the histograms
bin_edges = np.linspace(-5, 8, 15)

# Define the arguments dictionary
ARGS = {
    "samples1": samples1,
    "samples2": samples2,
    "bin_edges": bin_edges,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
combine_and_save Ran
ran on a synthesized input
jayaram-r/adversarial-detection/expts/generate_samples_custom.py
code served (permissive licence) · get_code("6d6b0a89408bd010")
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. 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 os
import numpy as np
import tempfile

# Create a temporary directory for saving
temp_dir = tempfile.mkdtemp()

# Construct the arguments
save_path = temp_dir

# Create small sample data for adversarial examples
# Lists of batches that will be concatenated
data_adver = [
    np.random.randn(2, 4, 3).astype(np.float32),  # batch 1: 2 samples, 4x3 features
    np.random.randn(2, 4, 3).astype(np.float32),  # batch 2: 2 samples, 4x3 features
]

labels_adver = [
    np.array([0, 1], dtype=np.int64),  # batch 1 labels
    np.array([2, 0], dtype=np.int64),  # batch 2 labels
]

# Create small sample data for clean examples
data_clean = [
    np.random.randn(2, 4, 3).astype(np.float32),  # batch 1: 2 samples, 4x3 features
    np.random.randn(2, 4, 3).astype(np.float32),  # batch 2: 2 samples, 4x3 features
]

labels_clean = [
    np.array([1, 2], dtype=np.int64),  # batch 1 labels
    np.array([0, 1], dtype=np.int64),  # batch 2 labels
]

# Perturbation norms for each adversarial example
norm_perturb = [
    np.array([0.1, 0.2], dtype=np.float32),  # batch 1 norms
    np.array([0.15, 0.25], dtype=np.float32),  # batch 2 norms
]

# Whether each example was correctly classified
is_correct = [
    np.array([True, False], dtype=bool),  # batch 1
    np.array([True, True], dtype=bool),  # batch 2
]

# Whether each example is adversarial
is_adver = [
    np.array([True, True], dtype=bool),  # batch 1
    np.array([True, False], dtype=bool),  # batch 2
]

# Test labels (for dtype reference)
labels_te = np.array([0, 1, 2, 0], dtype=np.int64)

ARGS = {
    'save_path': save_path,
    'data_adver': data_adver,
    'labels_adver': labels_adver,
    'data_clean': data_clean,
    'labels_clean': labels_clean,
    'norm_perturb': norm_perturb,
    'is_correct': is_correct,
    'is_adver': is_adver,
    'labels_te': labels_te,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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
log_transform_data Ran
ran on a synthesized input
jayaram-r/adversarial-detection/expts/helpers/density_model_layer_statistics.py
code served (permissive licence) · get_code("b72af4b9b2c9b75e")
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 ndarray, ndim 1, dtype family float, shape (6,) 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 numpy as np

# Construct valid arguments for log_transform_data
# The function expects a numpy array with non-negative values

# Create a simple numpy array with positive values
data_in = np.array([0.0, 0.001, 0.1, 1.0, 10.0, 100.0])

ARGS = {
    'data_in': data_in
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (6,) — shape (6,) 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: none by the harness; any seed is in the setup block
select_covar_types Ran
ran on a synthesized input
jayaram-r/adversarial-detection/expts/helpers/density_model_layer_statistics.py
code served (permissive licence) · get_code("0c9e4fefa0d08c76")
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 list, 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
import numpy as np

# The function takes two simple integer parameters: nd (data dimension) and ns (number of samples)
# We'll test with a case that exercises one of the branches

# Test case 1: nd=15 (<=20), ns=200 (>=10*15=150)
# This should return ['full', 'tied']
nd = 15
ns = 200

ARGS = {
    'nd': nd,
    'ns': ns
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list 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

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 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

Detecting anomalous inputs, such as adversarial and out-of-distribution (OOD) inputs, is critical for classifiers (including deep neural networks or DNNs) deployed in real-world applications. While prior works have proposed various methods to detect such anomalous samples using information from the internal layer representations of a DNN, there is a lack of consensus on a principled approach for the different components of such a detection method. As a result, often heuristic and one-off methods are applied for different aspects of this problem. We propose an unsupervised anomaly detection framework based on the internal DNN layer representations in the form of a meta-algorithm with configurable components. We proceed to propose specific instantiations for each component of the meta-algorithm based on ideas grounded in statistical testing and anomaly detection. We evaluate the proposed methods on well-known image classification datasets with strong adversarial attacks and OOD inputs, including an adaptive attack that uses the internal layer representations of the DNN (often not considered in prior work). Comparisons with five recently-proposed competing detection methods demonstrates the effectiveness of our method in detecting adversarial and OOD inputs.

For agents

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

get_harvested_code_for_paper("2007.15147")
get_code_for_paper("2007.15147")
have("2007.15147")

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

curl https://syntology.ai/api/ran/2007.15147.json

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

[![Syntology run record](https://syntology.ai/api/ran/2007.15147.svg)](https://syntology.ai/paper/2007.15147)

Connect an agent — have() is free.