SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2202.08836 · ICML · 2022

Data-SUITE: Data-centric identification of in-distribution incongruous examples

Mihaela Van Der Schaar, Jonathan Crabbé, Nabeel Seedat

arXiv · PDF · Open in the Atlas

Code that ran

9 of 11 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 9 have no contract verdict; 0 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/2202.08836.json · badge: /api/ran/2202.08836.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
seedatnabeel/data-suite canonical (classifier label)
classifier basis not recorded on the link
9 ran · 2 no recorded run
FunctionStatusWhere it lives
all_equal2 Ran
ran on a synthesized input
seedatnabeel/data-suite/src/models/mcd.py
code served (permissive licence) · get_code("f7ce604d7f4dcace")
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 bool, ndim 0, dtype family float, shape () 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 numpy as np

def all_equal2(iterator):
    return len(set(iterator)) <= 1

# Test case 1: All equal elements
iterator_all_equal = [1, 1, 1, 1]

ARGS = {
    "iterator": iterator_all_equal
}
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 bool, ndim 0, dtype family float, shape () — shape () 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
compute_representation Ran
ran on a synthesized input
seedatnabeel/data-suite/src/models/representation.py
code served (permissive licence) · get_code("7357cd13309d3f31")
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 numpy as np
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler
from tensorflow import keras
from tensorflow.keras import layers

np.random.seed(42)

# Create small synthetic datasets with distinct dimensions
train_data = np.random.randn(5, 4).astype(np.float32)  # 5 samples, 4 features
test_data = np.random.randn(3, 4).astype(np.float32)   # 3 samples, 4 features
copula_samples = np.random.randn(4, 4).astype(np.float32)  # 4 samples, 4 features

ARGS = {
    'train': train_data,
    'test': test_data,
    'copula_samples': copula_samples,
    'n_components': 2,
    'rep_type': 'pca',
    'seed': 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, tensorflow — 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
covariance_comparison Ran
ran on a synthesized input
seedatnabeel/data-suite/src/utils/data_utils.py
code served (permissive licence) · get_code("9d9f5093936c406a")
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 numpy as np
import matplotlib
matplotlib.use('Agg')  # Non-interactive backend to avoid display issues

# Create small test arrays
np.random.seed(0)

# Create clean array: shape (6, 4) - 6 samples, 4 features
clean_array = np.random.randn(6, 4).astype(np.float32)

# Create noisy array: shape (6, 4) - same shape as clean array
# Add some noise to make it distinct
noisy_array = clean_array + np.random.randn(6, 4) * 0.3

ARGS = {
    "clean_array": clean_array,
    "noisy_array": noisy_array,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, sklearn — 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
inlier_outlier_dicts Ran
ran on a synthesized input
seedatnabeel/data-suite/src/utils/helpers.py
code served (permissive licence) · get_code("3f0f7bcbd34e50f4")
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 pandas as pd
import numpy as np
from copy import deepcopy

# Create sample conformal prediction dataframes
conformal_dict = {}

# Feature 0
df0 = pd.DataFrame({
    'true_val': [1.0, 2.0, 10.0, 3.5],
    'min': [0.8, 1.9, 2.0, 3.2],
    'max': [1.2, 2.1, 3.0, 3.8],
    'conf_interval': [0.4, 0.2, 1.0, 0.6]
})
conformal_dict[0] = df0

# Feature 1
df1 = pd.DataFrame({
    'true_val': [5.0, 6.0, 20.0, 7.5],
    'min': [4.9, 5.8, 8.0, 7.2],
    'max': [5.1, 6.2, 9.0, 7.8],
    'conf_interval': [0.2, 0.4, 1.0, 0.6]
})
conformal_dict[1] = df1

# Suspect features to check
suspect_features = [0, 1]

ARGS = {
    'conformal_dict': conformal_dict,
    'suspect_features': suspect_features
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
representation_class_based Ran
ran on a synthesized input
seedatnabeel/data-suite/src/models/representation.py
code served (permissive licence) · get_code("8515fb9ae1ae6e7b")
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 numpy as np
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler

# Create small sample data for train and copula_samples
np.random.seed(42)
train_data = np.random.randn(5, 4).astype(np.float32)  # 5 samples, 4 features
copula_data = np.random.randn(3, 4).astype(np.float32)  # 3 samples, 4 features

ARGS = {
    'train': train_data,
    'copula_samples': copula_data,
    'n_components': 2,
    'rep_type': 'pca',
    'seed': 42
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — 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
return_diagonal Ran
ran on a synthesized input
seedatnabeel/data-suite/src/utils/data_utils.py
code served (permissive licence) · get_code("8827b14a2d884cb0")
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
import numpy as np

torch.manual_seed(0)

# Create a 2D array (using numpy for this function)
array = np.array([[1.0, 2.0, 3.0],
                  [4.0, 5.0, 6.0],
                  [7.0, 8.0, 9.0]], dtype=np.float32)

ARGS = {
    "array": array,
    "just_diag": True
}
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 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
scaler Ran
ran on a synthesized input
seedatnabeel/data-suite/src/utils/data_utils.py
code served (permissive licence) · get_code("db182a4e5b31bf62")
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 numpy as np

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

# Create small foreground and background data arrays
# fg: shape (2, 4) - 2 samples, 4 features
fg = np.random.randn(2, 4).astype(np.float32)

# bg: shape (3, 4) - 3 samples, 4 features (same feature dimension as fg)
bg = np.random.randn(3, 4).astype(np.float32)

# Define the arguments dictionary
ARGS = {
    'fg': fg,
    'bg': bg,
    'center': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — 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
sort_cis_all Ran
ran on a synthesized input
seedatnabeel/data-suite/src/utils/helpers.py
code served (permissive licence) · get_code("58c9f9694a3b2975")
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 pandas as pd
import numpy as np

# Create sample DataFrames for conformal_dict
# Each DataFrame needs a 'norm_interval' column and an index
np.random.seed(0)

# Create conformal_dict with DataFrames for different features
conformal_dict = {}
for feature in ['feature_1', 'feature_2', 'feature_3']:
    # Create a DataFrame with 10 rows and a 'norm_interval' column
    data = {
        'norm_interval': np.random.uniform(0.1, 5.0, 10),
        'value': np.random.randn(10)
    }
    conformal_dict[feature] = pd.DataFrame(data, index=range(10))

# Create inliers_dict mapping features to arrays of inlier indices
inliers_dict = {}
for feature in ['feature_1', 'feature_2', 'feature_3']:
    # Select some indices as inliers (e.g., indices 1-7)
    inliers_dict[feature] = np.array([1, 2, 3, 4, 5, 6, 7])

# Create suspect_features list
suspect_features = ['feature_1', 'feature_2']

ARGS = {
    'conformal_dict': conformal_dict,
    'inliers_dict': inliers_dict,
    'suspect_features': suspect_features
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
sort_cis_synth Ran
ran on a synthesized input
seedatnabeel/data-suite/src/utils/helpers.py
code served (permissive licence) · get_code("53035afb7a99185e")
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 pandas as pd
import numpy as np

# Create sample conformal intervals dataframes
np.random.seed(0)

n_samples = 20
n_features = 3

# Create conformal dataframes for each feature
conformal_dict = {}
for i in range(n_features):
    feature_name = f'feature_{i}'
    df = pd.DataFrame({
        'norm_interval': np.random.uniform(0.1, 2.0, n_samples),
        'lower_bound': np.random.uniform(-1, 0, n_samples),
        'upper_bound': np.random.uniform(0, 1, n_samples)
    })
    conformal_dict[feature_name] = df

# Create inliers dictionaries - select random subset of indices as inliers
inliers_dict = {}
for i in range(n_features):
    feature_name = f'feature_{i}'
    # Select 80% of indices as inliers
    inlier_indices = np.random.choice(n_samples, size=int(0.8 * n_samples), replace=False)
    inliers_dict[feature_name] = inlier_indices

# Create suspect features list
suspect_features = ['feature_0', 'feature_1', 'feature_2']

# Create proportion parameter
proportion = 0.1

ARGS = {
    'conformal_dict': conformal_dict,
    'inliers_dict': inliers_dict,
    'suspect_features': suspect_features,
    'proportion': proportion
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
get_mse_table Not yet run
unverified (harvested, no recorded run)
seedatnabeel/data-suite/src/utils/graphics.py
code served (permissive licence) · get_code("34f7beadd5fda2a1")
How it ran

no invocation recorded.

sample_copula Not yet run
unverified (harvested, no recorded run)
seedatnabeel/data-suite/src/models/benchmarks.py
code served (permissive licence) · get_code("ab6621513840038b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 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 11 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

Systematic quantification of data quality is critical for consistent model performance. Prior works have focused on out-of-distribution data. Instead, we tackle an understudied yet equally important problem of characterizing incongruous regions of in-distribution (ID) data, which may arise from feature space heterogeneity. To this end, we propose a paradigm shift with Data-SUITE: a datacentric AI framework to identify these regions, independent of a task-specific model. Data-SUITE leverages copula modeling, representation learning, and conformal prediction to build featurewise confidence interval estimators based on a set of training instances. These estimators can be used to evaluate the congruence of test instances with respect to the training set, to answer two practically useful questions: (1) which test instances will be reliably predicted by a model trained with the training instances? and (2) can we identify incongruous regions of the feature space so that data owners understand the data's limitations or guide future data collection? We empirically validate Data-SUITE's performance and coverage guarantees and demonstrate on cross-site medical data, biased data, and data with concept drift, that Data-SUITE best identifies ID regions where a downstream model may be reliable (independent of said model). We also illustrate how these identified regions can provide insights into datasets and highlight their limitations.

For agents

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

get_harvested_code_for_paper("2202.08836")
get_code_for_paper("2202.08836")
have("2202.08836")

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

curl https://syntology.ai/api/ran/2202.08836.json

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

[![Syntology run record](https://syntology.ai/api/ran/2202.08836.svg)](https://syntology.ai/paper/2202.08836)

Connect an agent — have() is free.