SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2110.09599 · ICML · 2022

Label-Descriptive Patterns and Their Application to Characterizing Classification Errors

Michael Hedderich, Dietrich Klakow, Jilles Vreeken, Jonas Fischer

arXiv · PDF · Open in the Atlas

Code that ran

3 of 5 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (3 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 3 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/2110.09599.json · badge: /api/ran/2110.09599.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
uds-lsv/pypremise community
Papers with Code archive
3 ran · 2 no recorded run
FunctionStatusWhere it lives
Pattern Ran
ran on a synthesized input · constructed an object; no result was computed
uds-lsv/pypremise/src/pypremise/core.py
code served (permissive licence) · get_code("bf04d577bbf2ab0a")
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, Pattern: 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
from typing import List, Union

class Pattern:
    """
    A pattern is a conjunction of disjunctive items,
    e.g. (1 or 2) and (3) and (4 or 5 or 6)
    """

    def __init__(self, clauses: List[List[Union[int, str]]]):
        self.clauses: List[List[Union[int, str]]] = clauses

    def __str__(self):
        representation = ' and '.join(['(' + '-or-'.join([str(item) for item in clause]) + ')' for clause in self.clauses])
        return f'{representation}'


ARGS = {
    'clauses': [
        [1, 2],
        [3],
        [4, 5, 6]
    ]
}
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 Pattern — 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 Pattern 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
Premise Ran
ran on a synthesized input · constructed an object; no result was computed
uds-lsv/pypremise/src/pypremise/core.py
code served (permissive licence) · get_code("cc1fe08bf9887a8a")
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, Premise: 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 tempfile
import os
from typing import Mapping, Optional, List

# Define the ARGS dictionary with all parameters for Premise.__init__

# Create minimal vocabulary and embedding
voc_index_to_token = {
    0: "word0",
    1: "word1",
    2: "word2",
    3: "word3",
}

# Create minimal embedding vectors (4 features, 3-dimensional embeddings)
embedding_index_to_vector = {
    0: [0.1, 0.2, 0.3],
    1: [0.4, 0.5, 0.6],
    2: [0.7, 0.8, 0.9],
    3: [1.0, 1.1, 1.2],
}

ARGS = {
    "voc_index_to_token": voc_index_to_token,
    "embedding_index_to_vector": embedding_index_to_vector,
    "embedding_dimensionality": 3,
    "max_neighbor_distance": 2,
    "fisher_p_value": 0.01,
    "clause_max_overlap": 0.05,
    "min_overlap": 0.3,
    "group_0_name": "control",
    "group_1_name": "treatment",
    "premise_engine": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pypremise — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Premise — 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 Premise 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
PremiseResult Ran
ran on a synthesized input · constructed an object; no result was computed
uds-lsv/pypremise/src/pypremise/core.py
code served (permissive licence) · get_code("f68c9d8148d1c0c4")
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, PremiseResult: 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
from enum import Enum
from typing import List, Union

class Direction(Enum):
    GROUP_0 = 0
    GROUP_1 = 1

class Pattern:
    """
    A pattern is a conjunction of disjunctive items,
    e.g. (1 or 2) and (3) and (4 or 5 or 6)
    """

    def __init__(self, clauses: List[List[Union[int, str]]]):
        self.clauses: List[List[Union[int, str]]] = clauses

    def __str__(self):
        representation = ' and '.join(['(' + '-or-'.join([str(item) for item in clause]) + ')' for clause in self.clauses])
        return f'{representation}'

# Create test values
pattern = Pattern([[1, 2], [3], [4, 5, 6]])
direction = Direction.GROUP_0
label0_count = 42
label1_count = 58
mdl_gain = 0.85
fisher_value = 3.14
group_0_name = "control"
group_1_name = "treatment"

ARGS = {
    "pattern": pattern,
    "direction": direction,
    "label0_count": label0_count,
    "label1_count": label1_count,
    "mdl_gain": mdl_gain,
    "fisher_value": fisher_value,
    "group_0_name": group_0_name,
    "group_1_name": group_1_name,
}
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 PremiseResult — 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 PremiseResult 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
Direction Not yet run
unverified (harvested, no recorded run)
uds-lsv/pypremise/src/pypremise/core.py
code served (permissive licence) · get_code("c610d635a6ef042e")
How it ran

no invocation recorded.

PremiseInstance Not yet run
unverified (harvested, no recorded run)
uds-lsv/pypremise/src/pypremise/core.py
code served (permissive licence) · get_code("6ab5be9a22760f72")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 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 5 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

State-of-the-art deep learning methods achieve human-like performance on many tasks, but make errors nevertheless. Characterizing these errors in easily interpretable terms gives insight into whether a classifier is prone to making systematic errors, but also gives a way to act and improve the classifier. We propose to discover those featurevalue combinations (i.e., patterns) that strongly correlate with correct resp. erroneous predictions to obtain a global and interpretable description for arbitrary classifiers. We show this is an instance of the more general label description problem, which we formulate in terms of the Minimum Description Length principle. To discover a good pattern set, we develop the efficient PREMISE algorithm. Through an extensive set of experiments we show it performs very well in practice on both synthetic and real-world data. Unlike existing solutions, it ably recovers ground truth patterns, even on highly imbalanced data over many features. Through two case studies on Visual Question Answering and Named Entity Recognition, we confirm that PREMISE gives clear and actionable insight into the systematic errors made by modern NLP classifiers.

For agents

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

get_harvested_code_for_paper("2110.09599")
get_code_for_paper("2110.09599")
have("2110.09599")

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

curl https://syntology.ai/api/ran/2110.09599.json

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

[![Syntology run record](https://syntology.ai/api/ran/2110.09599.svg)](https://syntology.ai/paper/2110.09599)

Connect an agent — have() is free.