SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1807.03653 · 2018

Handling Incomplete Heterogeneous Data using VAEs

arXiv · PDF · Open in the Atlas

Code that ran

4 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 4 have no contract verdict; 0 failures of our instrument; 7 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/1807.03653.json · badge: /api/ran/1807.03653.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
probabilistic-learning/HI-VAE canonical (classifier label)
classifier basis: census
4 ran · 7 no recorded run
FunctionStatusWhere it lives
getArgs Ran
ran on a synthesized input
probabilistic-learning/HI-VAE/parser_arguments.py
code served (permissive licence) · get_code("3692070b96d5bea1")
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 Namespace. 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 argparse

def getArgs(argv=None):
    parser = argparse.ArgumentParser(description='Default parameters of the models', formatter_class=argparse.ArgumentDefaultsHelpFormatter)
    parser.add_argument('--batch_size', type=int, default=200, help='Size of the batches')
    parser.add_argument('--epochs', type=int, default=5001, help='Number of epochs of the simulations')
    parser.add_argument('--perp', type=int, default=10, help='Perplexity for the t-SNE')
    parser.add_argument('--train', type=int, default=1, help='Training model flag')
    parser.add_argument('--display', type=int, default=1, help='Display option flag')
    parser.add_argument('--save', type=int, default=1000, help='Save variables every save iterations')
    parser.add_argument('--restore', type=int, default=0, help='To restore session, to keep training or evaluation')
    parser.add_argument('--plot', type=int, default=1, help='Plot results flag')
    parser.add_argument('--dim_latent_s', type=int, default=10, help='Dimension of the categorical space')
    parser.add_argument('--dim_latent_z', type=int, default=2, help='Dimension of the Z latent space')
    parser.add_argument('--dim_latent_y', type=int, default=10, help='Dimension of the Y latent space')
    parser.add_argument('--dim_latent_y_partition', type=int, nargs='+', help='Partition of the Y latent space')
    parser.add_argument('--miss_percentage_train', type=float, default=0.0, help='Percentage of missing data in training')
    parser.add_argument('--miss_percentage_test', type=float, default=0.0, help='Percentage of missing data in test')
    parser.add_argument('--model_name', type=str, default='model_new', help='File of the training model')
    parser.add_argument('--save_file', type=str, default='new_mnist_zdim5_ydim10_4images_', help='Save file name')
    parser.add_argument('--data_file', type=str, default='MNIST_data', help='File with the data')
    parser.add_argument('--types_file', type=str, default='mnist_train_types2.csv', help='File with the types of the data')
    parser.add_argument('--miss_file', type=str, default='Missing_test.csv', help='File with the missing indexes mask')
    parser.add_argument('--true_miss_file', type=str, help='File with the missing indexes when there are NaN in the data')
    return parser.parse_args(argv)

ARGS = {
    'argv': [
        '--batch_size', '32',
        '--epochs', '100',
        '--perp', '5',
        '--train', '1',
        '--display', '1',
        '--save', '50',
        '--restore', '0',
        '--plot', '1',
        '--dim_latent_s', '8',
        '--dim_latent_z', '4',
        '--dim_latent_y', '6',
        '--dim_latent_y_partition', '2', '4',
        '--miss_percentage_train', '0.1',
        '--miss_percentage_test', '0.05',
        '--model_name', 'test_model',
        '--save_file', 'test_save_',
        '--data_file', 'test_data',
        '--types_file', 'test_types.csv',
        '--miss_file', 'test_missing.csv',
        '--true_miss_file', 'test_true_missing.csv'
    ]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Namespace — a Namespace output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Namespace 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
next_batch Ran
ran on a synthesized input
probabilistic-learning/HI-VAE/read_functions.py
code served (permissive licence) · get_code("21e35880e258189a")
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)

# Construct arguments for next_batch function
data = torch.randn(8, 10)  # 8 samples, 10 total features
types_dict = [
    {'dim': '4'},  # First feature group has 4 dimensions
    {'dim': '3'},  # Second feature group has 3 dimensions
    {'dim': '3'},  # Third feature group has 3 dimensions
]
miss_mask = torch.randint(0, 2, (8, 10))  # 8 samples, 10 features (binary mask)
batch_size = 2
index_batch = 0

ARGS = {
    'data': data,
    'types_dict': types_dict,
    'miss_mask': miss_mask,
    'batch_size': batch_size,
    'index_batch': index_batch,
}
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
read_data Ran
ran on a synthesized input
probabilistic-learning/HI-VAE/read_functions.py
code served (permissive licence) · get_code("1b8f49db2dcb26be")
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 tempfile
import os
import csv
import numpy as np

# Create temporary files with valid CSV data
temp_dir = tempfile.mkdtemp()

# Create data file with numeric data (5 samples, 3 variables)
data_file = os.path.join(temp_dir, "data.csv")
with open(data_file, 'w', newline='') as f:
    writer = csv.writer(f)
    writer.writerow([1.0, 2.0, 3.5])
    writer.writerow([1.0, 2.0, 4.5])
    writer.writerow([2.0, 3.0, 5.5])
    writer.writerow([2.0, 3.0, 6.5])
    writer.writerow([1.0, 2.0, 7.5])

# Create types file describing variable types and dimensions
types_file = os.path.join(temp_dir, "types.csv")
with open(types_file, 'w', newline='') as f:
    writer = csv.writer(f)
    writer.writerow(['type', 'dim'])
    writer.writerow(['cat', '2'])
    writer.writerow(['ordinal', '3'])
    writer.writerow(['real', '1'])

# Create miss file (missing data indicator)
miss_file = os.path.join(temp_dir, "miss.csv")
with open(miss_file, 'w', newline='') as f:
    writer = csv.writer(f)
    writer.writerow([1, 1])
    writer.writerow([3, 2])

# Create true_miss_file (true missing positions)
true_miss_file = os.path.join(temp_dir, "true_miss.csv")
with open(true_miss_file, 'w', newline='') as f:
    writer = csv.writer(f)
    writer.writerow([2, 1])
    writer.writerow([4, 2])

ARGS = {
    'data_file': data_file,
    'types_file': types_file,
    'miss_file': miss_file,
    'true_miss_file': true_miss_file,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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
samples_concatenation Ran
ran on a synthesized input
probabilistic-learning/HI-VAE/read_functions.py
code served (permissive licence) · get_code("34a3b61666205c08")
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

# Construct valid arguments for samples_concatenation
# The function expects a list of batches, where each batch is a dict with keys: 'x', 'y', 'z', 's'

# Create first batch
batch_1 = {
    'x': [np.random.randn(2, 3), np.random.randn(2, 5)],  # list of arrays to concatenate along axis 1
    'y': np.random.randn(2, 4),
    'z': np.random.randn(2, 6),
    's': np.random.randn(2, 7)
}

# Create second batch
batch_2 = {
    'x': [np.random.randn(3, 3), np.random.randn(3, 5)],  # different batch size (3 vs 2)
    'y': np.random.randn(3, 4),
    'z': np.random.randn(3, 6),
    's': np.random.randn(3, 7)
}

# Create list of batches
samples = [batch_1, batch_2]

ARGS = {
    'samples': samples
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — 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
batch_normalization Not yet run
unverified (harvested, no recorded run)
probabilistic-learning/HI-VAE/VAE_functions.py
code served (permissive licence) · get_code("3d429bcfc3c677e8")
How it ran

no invocation recorded.

cost_function Not yet run
unverified (harvested, no recorded run)
probabilistic-learning/HI-VAE/model_HIVAE_factorized.py
code served (permissive licence) · get_code("a9d7aa7274a5abf9")
How it ran

no invocation recorded.

loglik_cat Not yet run
unverified (harvested, no recorded run)
probabilistic-learning/HI-VAE/loglik_models_missing_normalize.py
code served (permissive licence) · get_code("54df9b43971857da")
How it ran

no invocation recorded.

loglik_pos Not yet run
unverified (harvested, no recorded run)
probabilistic-learning/HI-VAE/loglik_models_missing_normalize.py
code served (permissive licence) · get_code("7e1aac4d69cb3072")
How it ran

no invocation recorded.

loglik_real Not yet run
unverified (harvested, no recorded run)
probabilistic-learning/HI-VAE/loglik_models_missing_normalize.py
code served (permissive licence) · get_code("fc6abe9dbfd369fb")
How it ran

no invocation recorded.

place_holder_types Not yet run
unverified (harvested, no recorded run)
probabilistic-learning/HI-VAE/VAE_functions.py
code served (permissive licence) · get_code("b4ed17fa0cc84438")
How it ran

no invocation recorded.

s_proposal_multinomial Not yet run
unverified (harvested, no recorded run)
probabilistic-learning/HI-VAE/VAE_functions.py
code served (permissive licence) · get_code("dc957f51387e37d9")
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, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 10 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
cost_function 1 other paper
1 later
2006.12871 (2020-06)
not-MIWAE: Deep Generative Modelling with Missing not at Random Data

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

Variational autoencoders (VAEs), as well as other generative models, have been shown to be efficient and accurate for capturing the latent structure of vast amounts of complex high-dimensional data. However, existing VAEs can still not directly handle data that are heterogenous (mixed continuous and discrete) or incomplete (with missing data at random), which is indeed common in real-world applications. In this paper, we propose a general framework to design VAEs suitable for fitting incomplete heterogenous data. The proposed HI-VAE includes likelihood models for real-valued, positive real valued, interval, categorical, ordinal and count data, and allows accurate estimation (and potentially imputation) of missing data. Furthermore, HI-VAE presents competitive predictive performance in supervised tasks, outperforming supervised models when trained on incomplete data.

For agents

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

get_harvested_code_for_paper("1807.03653")
get_code_for_paper("1807.03653")
have("1807.03653")

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

curl https://syntology.ai/api/ran/1807.03653.json

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

[![Syntology run record](https://syntology.ai/api/ran/1807.03653.svg)](https://syntology.ai/paper/1807.03653)

Connect an agent — have() is free.