SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1809.01093 · 2018

Adversarial Attacks on Node Embeddings via Graph Poisoning

arXiv · PDF · Open in the Atlas

Code that ran

5 of 8 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 5 have no contract verdict; 0 failures of our instrument; 3 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/1809.01093.json · badge: /api/ran/1809.01093.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
abojchevski/node_embedding_attack community (classifier label)
classifier basis not recorded on the link
5 ran · 3 no recorded run
FunctionStatusWhere it lives
estimate_delta_eigenvecs Ran
ran on a synthesized input
abojchevski/node_embedding_attack/node_embedding_attack/perturbation_attack.py
code served (permissive licence) · get_code("a050cc1712273038")
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 3, dtype family float, shape (2, 3, 4) 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

# Set random seed for reproducibility
np.random.seed(0)

# Define problem dimensions
n_nodes = 4  # number of nodes
n_candidates = 2  # number of candidate edge flips
n_eigenvecs = 3  # number of eigenvectors (k)

# candidates: shape [n_candidates, 2] - pairs of node indices for edge flips
candidates = np.array([[0, 1], [2, 3]], dtype=np.int64)

# flip_indicator: shape [n_candidates] - +1 for adding edge, -1 for removing edge
flip_indicator = np.array([1, -1], dtype=np.float64)

# degrees: shape [n_nodes] - degree of each node
degrees = np.array([2, 2, 1, 1], dtype=np.float64)

# vals_org: shape [n_eigenvecs] - eigenvalues of the clean graph
vals_org = np.array([3.5, 2.1, 0.8], dtype=np.float64)

# vecs_org: shape [n_nodes, n_eigenvecs] - eigenvectors of the clean graph
vecs_org = np.array([
    [0.5, 0.3, 0.1],
    [0.4, 0.6, 0.2],
    [0.3, 0.2, 0.8],
    [0.2, 0.5, 0.3]
], dtype=np.float64)

# delta_eigvals: shape [n_candidates, n_eigenvecs] - estimated change in eigenvalues
delta_eigvals = np.array([
    [0.1, 0.05, 0.02],
    [-0.08, -0.03, -0.01]
], dtype=np.float64)

# pinvs: shape [n_eigenvecs, n_nodes, n_nodes] - precomputed pseudo-inverse matrices
pinvs = np.zeros((n_eigenvecs, n_nodes, n_nodes), dtype=np.float64)
for k in range(n_eigenvecs):
    # Create a simple invertible matrix for each eigenvalue
    A = np.eye(n_nodes) + 0.1 * np.random.randn(n_nodes, n_nodes)
    pinvs[k] = np.linalg.pinv(A)

ARGS = {
    "candidates": candidates,
    "flip_indicator": flip_indicator,
    "degrees": degrees,
    "vals_org": vals_org,
    "vecs_org": vecs_org,
    "delta_eigvals": delta_eigvals,
    "pinvs": pinvs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 3, dtype family float, shape (2, 3, 4) — shape (2, 3, 4) 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
flip_candidates Ran
ran on a synthesized input
abojchevski/node_embedding_attack/node_embedding_attack/utils.py
code served (permissive licence) · get_code("a1c789daf14f4f81")
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 csr_matrix. 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 as np
import scipy.sparse as sp

# Create a small adjacency matrix (5x5)
n_nodes = 5
# Create a simple sparse adjacency matrix in CSR format
# Adding some edges: (0,1), (1,2), (2,3), (3,4)
row = np.array([0, 1, 2, 3])
col = np.array([1, 2, 3, 4])
data = np.array([1, 1, 1, 1])
adj_matrix = sp.csr_matrix((data, (row, col)), shape=(n_nodes, n_nodes))

# Create candidates array with shape [?, 2]
# These are edge pairs to flip
candidates = np.array([
    [0, 1],  # flip existing edge
    [0, 2],  # flip non-existing edge
    [1, 3],  # flip non-existing edge
])

ARGS = {
    "adj_matrix": adj_matrix,
    "candidates": candidates,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type csr_matrix — a csr_matrix output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a csr_matrix 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
generate_candidates_removal_minimum_spanning_tree Ran
ran on a synthesized input
abojchevski/node_embedding_attack/node_embedding_attack/utils.py
code served (permissive licence) · get_code("57a4b1e0c51e83d9")
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 2, dtype family int, shape (5, 2) 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
import scipy.sparse as sp

# Create a small sparse adjacency matrix for a graph
# We'll create a 5-node graph with some edges
n_nodes = 5
# Create a simple graph structure using COO format, then convert to CSR
row = np.array([0, 0, 1, 1, 2, 2, 3, 3, 4])
col = np.array([1, 2, 2, 3, 3, 4, 4, 0, 0])
data = np.ones(len(row))

# Create symmetric adjacency matrix (undirected graph)
adj_matrix = sp.coo_matrix((data, (row, col)), shape=(n_nodes, n_nodes))
adj_matrix = adj_matrix + adj_matrix.T  # Make symmetric
adj_matrix = adj_matrix.tocsr()

ARGS = {
    "adj_matrix": adj_matrix
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family int, shape (5, 2) — shape (5, 2) 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
sample_random_walks Ran
ran on a synthesized input
abojchevski/node_embedding_attack/node_embedding_attack/embedding.py
code served (permissive licence) · get_code("25754f382a10e3dc")
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 2, dtype family int, shape (10, 3) 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
import scipy.sparse as sp

# Create a small sparse adjacency matrix for a simple graph
# 5 nodes with some edges
n_nodes = 5
rows = np.array([0, 0, 1, 1, 2, 3, 4])
cols = np.array([1, 2, 0, 3, 1, 2, 0])
data = np.array([1, 1, 1, 1, 1, 1, 1])
adj_matrix = sp.csr_matrix((data, (rows, cols)), shape=(n_nodes, n_nodes))

ARGS = {
    "adj_matrix": adj_matrix,
    "walk_length": 3,
    "walks_per_node": 2,
    "seed": 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family int, shape (10, 3) — shape (10, 3) 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
sum_of_powers Ran
ran on a synthesized input
abojchevski/node_embedding_attack/node_embedding_attack/utils.py
code served (permissive licence) · get_code("2011e8002a39184b")
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 (4,) 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
import numba

# Construct arguments for sum_of_powers
x = np.array([1.0, 2.0, 3.0, 4.0], dtype=np.float64)
power = 3

ARGS = {
    "x": x,
    "power": power
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, 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 (4,) — shape (4,) 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
deepwalk_svd Not yet run
unverified (harvested, no recorded run)
abojchevski/node_embedding_attack/node_embedding_attack/embedding.py
code served (permissive licence) · get_code("b9b3772933c8c723")
How it ran

no invocation recorded.

estimate_loss_with_delta_eigenvals Not yet run
unverified (harvested, no recorded run)
abojchevski/node_embedding_attack/node_embedding_attack/perturbation_attack.py
code served (permissive licence) · get_code("88afa46c6f6f0129")
How it ran

no invocation recorded.

perturbation_top_flips Not yet run
unverified (harvested, no recorded run)
abojchevski/node_embedding_attack/node_embedding_attack/perturbation_attack.py
code served (permissive licence) · get_code("539907e904f02bd2")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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 8 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

The goal of network representation learning is to learn low-dimensional node embeddings that capture the graph structure and are useful for solving downstream tasks. However, despite the proliferation of such methods, there is currently no study of their robustness to adversarial attacks. We provide the first adversarial vulnerability analysis on the widely used family of methods based on random walks. We derive efficient adversarial perturbations that poison the network structure and have a negative effect on both the quality of the embeddings and the downstream tasks. We further show that our attacks are transferable since they generalize to many models and are successful even when the attacker is restricted.

For agents

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

get_harvested_code_for_paper("1809.01093")
get_code_for_paper("1809.01093")
have("1809.01093")

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

curl https://syntology.ai/api/ran/1809.01093.json

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

[![Syntology run record](https://syntology.ai/api/ran/1809.01093.svg)](https://syntology.ai/paper/1809.01093)

Connect an agent — have() is free.