SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2406.18865 · ICML · 2024

From Biased Selective Labels to Pseudo-Labels: An Expectation-Maximization Framework for Learning from Biased Decisions

Jenna Wiens, Trenton Chang

arXiv · PDF · Open in the Atlas

Code that ran

5 of 9 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; 4 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 1 failure of our instrument; 4 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/2406.18865.json · badge: /api/ran/2406.18865.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
mld3/disparate_censorship canonical (classifier label)
classifier basis: classifier_2025
3 ran · 1 no recorded run
mld3/dcem canonical (classifier label)
classifier basis: classifier_2025
2 ran · 3 no recorded run
FunctionStatusWhere it lives
_query_tensor_dataset Ran
ran, honoured the contract we drafted
mld3/dcem/dcem.py
code served (permissive licence) · get_code("a1027eae19d8ee87")
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) and returned Tensor, ndim 2, dtype family float, shape (2, 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 torch
from torch.utils.data import TensorDataset, Subset

torch.manual_seed(0)

# Create small tensors with distinct dimensions for each variable type
# TENSOR_INDICES = {'X': 0, 'A': 1, 'T': 2, 'Y_obs': 3, 'Y': 4, 'prop_score': 5}
X = torch.randn(2, 4)      # batch=2, dim=4
A = torch.randn(2, 3)      # batch=2, dim=3
T = torch.randn(2, 5)      # batch=2, dim=5
Y_obs = torch.randn(2, 6)  # batch=2, dim=6
Y = torch.randn(2, 7)      # batch=2, dim=7
prop_score = torch.randn(2, 8)  # batch=2, dim=8

# Create a TensorDataset with all tensors
tensor_dataset = TensorDataset(X, A, T, Y_obs, Y, prop_score)

# Test case 1: Direct TensorDataset access
dataset_1 = tensor_dataset
key_1 = 'X'

# Test case 2: Subset access
subset = Subset(tensor_dataset, indices=[0, 1])
dataset_2 = subset
key_2 = 'A'

# Use the first test case for ARGS
ARGS = {
    'dataset': dataset_1,
    'key': key_1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — 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 (2, 4) — shape (2, 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
causal_regularization_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
mld3/dcem/dcem.py
code served (permissive licence) · get_code("91f202f0c63eada3")
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) and returned Tensor, 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 torch.nn as nn
import numpy as np

torch.manual_seed(0)

# Create a simple loss function (CrossEntropyLoss)
loss_fn = nn.CrossEntropyLoss(reduction='none')

# Create tensors with appropriate shapes and values
batch_size = 2
num_classes = 2

# y_obs: observed labels (batch_size,) with values in {0, 1}
y_obs = torch.tensor([0, 1], dtype=torch.long)

# y_post: E-step estimates, probabilities in [0, 1] (batch_size,)
y_post = torch.tensor([0.3, 0.7], dtype=torch.float32)

# y_est: logits for outcome predictor P(Y|X) (batch_size, num_classes)
y_est = torch.randn(batch_size, num_classes, dtype=torch.float32)

# t_est: logits for propensity model P(T|X,A) (batch_size, num_classes)
t_est = torch.randn(batch_size, num_classes, dtype=torch.float32)

# reg: regularization parameter
reg = 1.0

ARGS = {
    'loss_fn': loss_fn,
    'y_obs': y_obs,
    'y_post': y_post,
    'y_est': y_est,
    't_est': t_est,
    'reg': reg
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, 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
cell_prototype Ran
ran, honoured the contract we drafted
mld3/disparate_censorship/simulation.py
pointer only (licence: NONE) · get_code("189ea404412bc405")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, mld3/disparate_censorship when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

down_staircase Ran
ran, honoured the contract we drafted
mld3/disparate_censorship/simulation.py
pointer only (licence: NONE) · get_code("faaaff7c5c616717")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, mld3/disparate_censorship when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_rotation_matrix Ran
ran, honoured the contract we drafted
mld3/disparate_censorship/simulation.py
pointer only (licence: NONE) · get_code("f8f1707bed2b4836")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, mld3/disparate_censorship when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

DCEM Not yet run
unverified (harvested, no recorded run)
mld3/dcem/dcem.py
code served (permissive licence) · get_code("1ec56f4886878dbc")
How it ran

no invocation recorded.

_enforce_bounds Not yet run
unverified (harvested, no recorded run)
mld3/dcem/dcem.py
code served (permissive licence) · get_code("e94e16d1e81731ac")
How it ran

no invocation recorded.

_enforce_scalar_bounds Not yet run
unverified (harvested, no recorded run)
mld3/dcem/dcem.py
code served (permissive licence) · get_code("498c019becd2246d")
How it ran

no invocation recorded.

parallel_call Not yet run
unverified (harvested, no recorded run)
mld3/disparate_censorship/experiment.py
pointer only (licence: NONE) · get_code("49676b7aed46f2cc")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, mld3/disparate_censorship when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 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 9 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

Selective labels occur when label observations are subject to a decision-making process; e.g., diagnoses that depend on the administration of laboratory tests. We study a clinically-inspired selective label problem called disparate censorship, where labeling biases vary across subgroups and unlabeled individuals are imputed as "negative" (i.e., no diagnostic test = no illness). Machine learning models naïvely trained on such labels could amplify labeling bias. Inspired by causal models of selective labels, we propose Disparate Censorship Expectation-Maximization (DCEM), an algorithm for learning in the presence of disparate censorship. We theoretically analyze how DCEM mitigates the effects of disparate censorship on model performance. We validate DCEM on synthetic data, showing that it improves bias mitigation (area between ROC curves) without sacrificing discriminative performance (AUC) compared to baselines. We achieve similar results in a sepsis classification task using clinical data.

For agents

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

get_harvested_code_for_paper("2406.18865")
get_code_for_paper("2406.18865")
have("2406.18865")

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

curl https://syntology.ai/api/ran/2406.18865.json

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

[![Syntology run record](https://syntology.ai/api/ran/2406.18865.svg)](https://syntology.ai/paper/2406.18865)

Connect an agent — have() is free.