SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2410.03396 · NeurIPS · 2024

GraphCroc: Cross-Correlation Autoencoder for Graph Structural Reconstruction

Shijin Duan, Ruyi Ding

arXiv · PDF · Open in the Atlas

Code that ran

11 of 14 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (6 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 6 have no contract verdict; 3 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/2410.03396.json · badge: /api/ran/2410.03396.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
sjduan/graphcroc canonical (classifier label)
classifier basis: classifier_2025
7 ran · 2 no recorded run
sjduan/GraphCroc canonical (classifier label)
classifier basis: census
4 ran · 1 no recorded run
FunctionStatusWhere it lives
Decoder Ran
ran on a synthesized input · constructed an object; no result was computed
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("34543d577c83afd2")
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 an instance of its own class, Decoder: the run constructed the object and did not call it, so no result was computed. 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)

# Define the supporting classes first
class GCN(nn.Module):
    def __init__(self, in_dim, out_dim, act, p):
        super(GCN, self).__init__()
        self.proj = nn.Linear(in_dim, out_dim)
        self.act = act
        self.drop = nn.Dropout(p=p) if p > 0.0 else nn.Identity()

    def forward(self, g, h):
        h = self.drop(h)
        h = torch.matmul(g, h)
        h = self.proj(h)
        h = self.act(h)
        return h

class Unpool(nn.Module):
    def __init__(self, *args):
        super(Unpool, self).__init__()

    def forward(self, g, h, idx):
        new_h = h.new_zeros([g.shape[0], h.shape[1]])
        new_h[idx] = h
        return (g, new_h)

def norm_g(g):
    degrees = torch.sum(g, dim=-1)
    degrees = torch.where(degrees == 0, torch.ones_like(degrees), degrees)
    g = g / degrees
    return g

class Decoder(nn.Module):
    def __init__(self, ks, dim, act, drop_p) -> None:
        super(Decoder, self).__init__()
        self.inp_LNs = nn.ModuleList()
        self.unpools = nn.ModuleList()
        self.up_gcns = nn.ModuleList()
        self.LNs = nn.ModuleList()
        self.l_n = len(ks)
        for i in range(self.l_n):
            self.inp_LNs.append(nn.LayerNorm(dim))
            self.unpools.append(Unpool())
            self.up_gcns.append(GCN(dim, dim, act, drop_p))
            self.LNs.append(nn.LayerNorm(dim))
        self.out_ln = nn.LayerNorm(dim)

    def forward(self, h, ori_h, down_outs, adj_ms, indices_list):
        for i in range(self.l_n):
            up_idx = self.l_n - i - 1
            g, idx = (adj_ms[up_idx], indices_list[up_idx])
            g, h = self.unpools[i](g, h, idx)
            h1 = self.inp_LNs[i](down_outs[up_idx] + h)
            g = norm_g(g)
            h = self.up_gcns[i](g, h1)
            h = self.LNs[i](h + h1)
        h = self.out_ln(h + ori_h)
        return h

# Create arguments for Decoder.__init__
ks = [2, 2]  # list of kernel sizes with length 2
dim = 4  # embedding dimension
act = nn.ReLU()  # activation function
drop_p = 0.1  # dropout probability

# For forward pass, we need:
# h: current hidden state (batch_size, dim)
# ori_h: original hidden state (batch_size, dim)
# down_outs: list of down outputs from encoder (one per layer)
# adj_ms: list of adjacency matrices (one per layer)
# indices_list: list of indices for unpooling (one per layer)

batch_size = 2
h = torch.randn(2, 4)  # smaller batch for unpooling
ori_h = torch.randn(2, 4)
down_outs = [torch.randn(2, 4), torch.randn(2, 4)]
adj_ms = [torch.randn(2, 2), torch.randn(2, 2)]
indices_list = [torch.tensor([0, 1]), torch.tensor([0, 1])]

ARGS = {
    'ks': ks,
    'dim': dim,
    'act': act,
    'drop_p': drop_p
}
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 Decoder — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Decoder 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
Encoder Ran
ran on a synthesized input · constructed an object; no result was computed
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("302c7028eb711ed5")
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 an instance of its own class, Encoder: the run constructed the object and did not call it, so no result was computed. 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)

# Small activation function
act = nn.ReLU()

# Parameters for Encoder
ks = [0.5, 0.5]  # pooling ratios for 2 layers
dim = 4  # feature dimension
drop_p = 0.1  # dropout probability

ARGS = {
    "ks": ks,
    "dim": dim,
    "act": act,
    "drop_p": drop_p,
}
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 Encoder — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Encoder 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
GCN Ran
ran on a synthesized input · constructed an object; no result was computed
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("9740f77d46c19ea2")
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 an instance of its own class, GCN: the run constructed the object and did not call it, so no result was computed. 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)

# GCN parameters:
# - in_dim: input dimension
# - out_dim: output dimension
# - act: activation function
# - p: dropout probability

in_dim = 4
out_dim = 3
act = nn.ReLU()
p = 0.5

# Create a small graph adjacency matrix (2x2) and feature matrix (2x4)
# g will be used in torch.matmul(g, h), so it needs compatible dimensions
g = torch.tensor([[1.0, 0.5], [0.5, 1.0]])  # 2x2 adjacency matrix
h = torch.randn(2, 4)  # 2x4 feature matrix (batch_size=2, in_dim=4)

ARGS = {
    "in_dim": in_dim,
    "out_dim": out_dim,
    "act": act,
    "p": p,
}
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 GCN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GCN 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
Pool Ran
ran on a synthesized input · constructed an object; no result was computed
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("855fd8cc32b38520")
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 an instance of its own class, Pool: the run constructed the object and did not call it, so no result was computed. 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)

# Create small test tensors
# g: adjacency matrix (num_nodes x num_nodes)
num_nodes = 5
g = torch.randint(0, 2, (num_nodes, num_nodes)).float()

# h: node features (num_nodes x in_dim)
in_dim = 4
h = torch.randn(num_nodes, in_dim)

# Parameters for Pool
k = 0.4  # fraction of nodes to keep
p = 0.1  # dropout probability

ARGS = {
    "k": k,
    "in_dim": in_dim,
    "p": p,
}
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 Pool — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Pool 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
Unet Ran
ran on a synthesized input · constructed an object; no result was computed
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("cfcfbfc18037a032")
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 an instance of its own class, Unet: the run constructed the object and did not call it, so no result was computed. 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)

class Args:
    def __init__(self):
        self.act = 'ReLU'
        self.mask_ratio = 0.1
        self.dim = 4
        self.drop_p = 0.1
        self.ks = [0.5, 0.5]

args = Args()

in_dim = 3
s_gcn_state = None
encoder_state = None
s_ln_state = None

ARGS = {
    'in_dim': in_dim,
    'args': args,
    's_gcn_state': s_gcn_state,
    'encoder_state': encoder_state,
    's_ln_state': s_ln_state
}
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 Unet — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Unet 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
Unpool Ran
ran on a synthesized input · constructed an object; no result was computed
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("10fffd68f36dce6e")
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 an instance of its own class, Unpool: the run constructed the object and did not call it, so no result was computed. 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)

# Create small tensors for testing
# g: graph features tensor with shape [batch_size, feature_dim]
g = torch.randn(2, 4)

# h: hidden state tensor with shape [num_nodes, feature_dim]
h = torch.randn(3, 4)

# idx: index tensor used to scatter h values into new_h
# This should be indices that reference positions in the new_h tensor
idx = torch.tensor([0, 2, 1])

ARGS = {
}
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 Unpool — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Unpool 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
compress_labels Ran
ran, honoured the contract we drafted
sjduan/GraphCroc/IMDB_B/wl_test_util.py
code served (permissive licence) · get_code("5b3253c77e503073")
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 int; 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

torch.manual_seed(0)

# The function compress_labels takes a list/iterable of labels
# and maps them to sequential integers starting from 1
# It iterates through labels and checks membership, so any iterable works

labels = [
    "cat",
    "dog", 
    "cat",
    "bird",
    "dog",
    "cat"
]

ARGS = {
    "labels": labels
}
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 int — 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
gen_node_feature Ran
ran, honoured the contract we drafted
sjduan/GraphCroc/IMDB_B/reconstructor.py
code served (permissive licence) · get_code("f39ca97dcafbfae0")
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 Tensor, ndim 2, dtype family float, shape (3, 1) 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.functional as F
from torch_geometric.data import Data

torch.manual_seed(0)

# Create a minimal data object with edge_index and num_nodes
edge_index = torch.tensor([[0, 1, 1, 2], [1, 0, 2, 1]], dtype=torch.long)
num_nodes = 3

data = Data(edge_index=edge_index, num_nodes=num_nodes)

ARGS = {
    "data": data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — 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 (3, 1) — shape (3, 1) 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
initialize_labels Ran
ran; our contract draft was wrong, not the code
sjduan/GraphCroc/IMDB_B/wl_test_util.py
code served (permissive licence) · get_code("0d3836f3da91b94c")
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

torch.manual_seed(0)

# The function takes an adj_matrix and sums along dim=1, then converts to list
# adj_matrix should be a 2D tensor
adj_matrix = torch.tensor([[1.0, 2.0, 3.0],
                           [4.0, 5.0, 6.0]], dtype=torch.float32)

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
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
relabel Ran
ran; our contract draft was wrong, not the code
sjduan/GraphCroc/IMDB_B/wl_test_util.py
code served (permissive licence) · get_code("6b9e2a85aa8b8be6")
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. 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

torch.manual_seed(0)

# Create a small adjacency matrix (4x4)
# This represents a graph with 4 nodes
adj_matrix = torch.tensor([
    [0, 1, 0, 1],
    [1, 0, 1, 0],
    [0, 1, 0, 1],
    [1, 0, 1, 0]
], dtype=torch.float32)

# Create labels for each node
labels = [10, 20, 30, 40]

ARGS = {
    "adj_matrix": adj_matrix,
    "labels": labels
}
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 list — a list 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
top_k_graph Ran
ran; our contract draft was wrong, not the code
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("9367ba5047a4d8ac")
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 torch

torch.manual_seed(0)

# Small graph adjacency matrix: 3 nodes
g = torch.tensor([
    [1.0, 1.0, 0.0],
    [1.0, 0.0, 1.0],
    [0.0, 1.0, 1.0]
])

# Node features: 3 nodes, 4 features
h = torch.tensor([
    [1.0, 2.0, 3.0, 4.0],
    [5.0, 6.0, 7.0, 8.0],
    [9.0, 10.0, 11.0, 12.0]
])

# Scores for each node: 3 values
scores = torch.tensor([0.9, 0.7, 0.5])

# k as a fraction (0.5 means select top 50% of nodes)
k = 0.5

ARGS = {
    "scores": scores,
    "g": g,
    "h": h,
    "k": k
}
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 — 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
calculate_metrics Not yet run
unverified (harvested, no recorded run)
sjduan/graphcroc/IMDB_B/reconstructor.py
code served (permissive licence) · get_code("033a97925e132d68")
How it ran

no invocation recorded.

gen_node_feature Not yet run
unverified (harvested, no recorded run)
sjduan/GraphCroc/IMDB_B/gc_train.py
code served (permissive licence) · get_code("65de7ac45402b8c9")
How it ran

no invocation recorded.

norm_g Not yet run
unverified (harvested, no recorded run)
sjduan/graphcroc/GraphCroc/UNET.py
code served (permissive licence) · get_code("021ac680f9625cd7")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 13 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
top_k_graph
brick of this paper's assembly: helper
1 other paper
1 dated earlier
1905.05178 (2019-05)
Graph U-Nets

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

Graph-structured data is integral to many applications, prompting the development of various graph representation methods. Graph autoencoders (GAEs), in particular, reconstruct graph structures from node embeddings. Current GAE models primarily utilize self-correlation to represent graph structures and focus on node-level tasks, often overlooking multi-graph scenarios. Our theoretical analysis indicates that selfcorrelation generally falls short in accurately representing specific graph features such as islands, symmetrical structures, and directional edges, particularly in smaller or multiple graph contexts. To address these limitations, we introduce a cross-correlation mechanism that significantly enhances the GAE representational capabilities. Additionally, we propose the GraphCroc, a new GAE that supports flexible encoder architectures tailored for various downstream tasks and ensures robust structural reconstruction, through a mirrored encoding-decoding process. This model also tackles the challenge of representation bias during optimization by implementing a loss-balancing strategy. Both theoretical analysis and numerical evaluations demonstrate that our methodology significantly outperforms existing self-correlation-based GAEs in graph structure reconstruction. Our code is available in https://github.com/sjduan/GraphCroc. * equal contribution Preprint. Under review.

For agents

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

get_harvested_code_for_paper("2410.03396")
get_code_for_paper("2410.03396")
have("2410.03396")

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

curl https://syntology.ai/api/ran/2410.03396.json

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

[![Syntology run record](https://syntology.ai/api/ran/2410.03396.svg)](https://syntology.ai/paper/2410.03396)

Connect an agent — have() is free.