SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2210.13647 · NeurIPS · 2022

Temporally Disentangled Representation Learning

Kun Zhang, Weiran Yao, Guangyi Chen

arXiv · PDF · Open in the Atlas

Code that ran

10 of 17 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (5 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 4 failures of our instrument; 7 no recorded run), measured 2026-11-22; 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/2210.13647.json · badge: /api/ran/2210.13647.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
weirayao/tdrl community
Papers with Code archive
10 ran · 7 no recorded run
FunctionStatusWhere it lives
BetaVAE_MLP Ran
ran on a synthesized input · constructed an object; no result was computed
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("de30ee3689394b09")
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, BetaVAE_MLP: 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)

ARGS = {
    "input_dim": 4,
    "z_dim": 5,
    "hidden_dim": 8,
}
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 BetaVAE_MLP — 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 BetaVAE_MLP 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
GroupLinearLayer Ran
ran on a synthesized input · constructed an object; no result was computed
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("3afb21c8fdb93144")
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, GroupLinearLayer: 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 GroupLinearLayer(nn.Module):
    """GroupLinearLayer computes N dinstinct linear transformations at once"""

    def __init__(self, din: int, dout: int, num_blocks: int, diagonal: bool=False, hidden=None) -> None:
        """Group Linear Layer module

        Args:
            din: The feature dimension of input data.
            dout: The projected dimensions of data.
            num_blocks: The number of linear transformation to compute at once.
            diagonal: Whether transition matrix is diagonal
        """
        super(GroupLinearLayer, self).__init__()
        assert hidden is None or type(hidden) == int
        self.hidden = hidden
        self.diagonal = diagonal
        assert (bool(self.hidden) and self.diagonal) == False
        if diagonal:
            self.d = nn.Parameter(0.01 * torch.randn(num_blocks, dout))
        elif hidden is None:
            self.w = nn.Parameter(0.01 * torch.randn(num_blocks, din, dout))
        else:
            assert isinstance(hidden, int)
            self.wh = nn.Parameter(0.01 * torch.randn(num_blocks, din, hidden))
            self.hw = nn.Parameter(0.01 * torch.randn(num_blocks, hidden, dout))

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        if self.diagonal:
            w = torch.diag_embed(self.d)
            x = x.permute(1, 0, 2)
            x = torch.bmm(x, w)
            x = x.permute(1, 0, 2)
        elif self.hidden is None:
            x = x.permute(1, 0, 2)
            x = torch.bmm(x, self.w)
            x = x.permute(1, 0, 2)
        else:
            x = x.permute(1, 0, 2)
            x = torch.bmm(x, self.wh)
            x = torch.bmm(x, self.hw)
            x = x.permute(1, 0, 2)
        return x

    def get_weight_matrix(self):
        if self.diagonal:
            return torch.diag_embed(self.d)
        elif self.hidden is None:
            return self.w
        else:
            return torch.matmul(self.wh, self.hw)


ARGS = {
    "din": 4,
    "dout": 3,
    "num_blocks": 2,
    "diagonal": False,
    "hidden": None
}
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 GroupLinearLayer — 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 GroupLinearLayer 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
MBDTransitionPrior Ran
ran on a synthesized input · constructed an object; no result was computed
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("99f55689f9c969b1")
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, MBDTransitionPrior: 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)

# Set up parameters for MBDTransitionPrior
lags = 3
latent_size = 4
bias = True

ARGS = {
    "lags": lags,
    "latent_size": latent_size,
    "bias": bias,
}
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 MBDTransitionPrior — 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 MBDTransitionPrior 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
NLayerLeakyMLP Ran
ran on a synthesized input · constructed an object; no result was computed
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("c4c2e657dc649aae")
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, NLayerLeakyMLP: 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)

ARGS = {
    'in_features': 4,
    'out_features': 3,
    'num_layers': 2,
    'hidden_dim': 64,
    'bias': True,
}
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 NLayerLeakyMLP — 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 NLayerLeakyMLP 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
NPChangeTransitionPrior Ran
ran on a synthesized input · constructed an object; no result was computed
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("59237073932bbf22")
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, NPChangeTransitionPrior: 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)

ARGS = {
    "lags": 2,
    "latent_size": 3,
    "embedding_dim": 4,
    "num_layers": 2,
    "hidden_dim": 8,
}
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 NPChangeTransitionPrior — 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 NPChangeTransitionPrior 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
_calculate_knots Ran
ran; our contract draft was wrong, not the code
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("09afbd64a67309fa")
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 torch.nn.functional as F

torch.manual_seed(0)

# Create a small tensor of bin lengths that sum to 1
# Shape: (2, 4) - batch size 2, 4 bins
lengths = torch.tensor([
    [0.1, 0.2, 0.3, 0.4],
    [0.25, 0.25, 0.25, 0.25]
], dtype=torch.float32)

lower = -1.0
upper = 1.0

ARGS = {
    "lengths": lengths,
    "lower": lower,
    "upper": upper
}
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
_monotonic_rational_spline Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("98886c9bc0d385ae")
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 3, 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
import torch.nn as nn
import torch.nn.functional as F

torch.manual_seed(0)

# Prepare tiny tensors for the function
batch_size = 2
num_bins = 4
input_dim = 3

# inputs: shape (batch_size, input_dim)
inputs = torch.randn(batch_size, input_dim) * 2.0

# widths: shape (batch_size, input_dim, num_bins)
widths = torch.rand(batch_size, input_dim, num_bins)
widths = widths / widths.sum(dim=-1, keepdim=True)  # Normalize to sum to 1

# heights: shape (batch_size, input_dim, num_bins)
heights = torch.rand(batch_size, input_dim, num_bins)
heights = heights / heights.sum(dim=-1, keepdim=True)  # Normalize to sum to 1

# derivatives: shape (batch_size, input_dim, num_bins - 1)
derivatives = torch.rand(batch_size, input_dim, num_bins - 1)

# lambdas: shape (batch_size, input_dim, num_bins) - optional, can be None
lambdas = torch.rand(batch_size, input_dim, num_bins)
lambdas = lambdas / lambdas.sum(dim=-1, keepdim=True)  # Normalize to sum to 1

ARGS = {
    'inputs': inputs,
    'widths': widths,
    'heights': heights,
    'derivatives': derivatives,
    'lambdas': lambdas,
    'inverse': False,
    'bound': 3.0,
    'min_bin_width': 0.001,
    'min_bin_height': 0.001,
    'min_derivative': 0.001,
    'min_lambda': 0.025,
    'eps': 1e-06,
}
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, ndim 3, 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
_searchsorted Ran
ran, honoured the contract we drafted
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("bb9c599734501473")
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 int, shape (2, 3) 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

torch.manual_seed(0)

# Create sorted_sequence: a 1D tensor of sorted values
sorted_sequence = torch.tensor([1.0, 3.0, 5.0, 7.0])

# Create values: a tensor of arbitrary shape containing values to search for
# Using shape (2, 3) to have distinct batch and feature dimensions
values = torch.tensor([
    [0.5, 2.0, 4.5],
    [6.0, 8.0, 3.5]
])

ARGS = {
    "sorted_sequence": sorted_sequence,
    "values": values
}
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 Tensor, ndim 2, dtype family int, shape (2, 3) — shape (2, 3) 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
_select_bins Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("48a69d3feffa4d35")
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 (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
import torch.nn as nn

torch.manual_seed(0)

# x should be a tensor with shape (context_batch_dims, input_dim, count_bins)
# We'll use shape (2, 4, 8) - batch_size=2, input_dim=4, count_bins=8
x = torch.randn(2, 4, 8)

# idx should be a tensor with shape (batch_dims, input_dim, 1)
# where batch_dims can broadcast over context_batch_dims
# We'll use shape (2, 4, 1) to match x's dimensions
idx = torch.randint(0, 8, (2, 4, 1))

ARGS = {
    "x": x,
    "idx": idx,
}
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 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
make_cost_matrix Ran
ran; our contract draft was wrong, not the code
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("60f32cea809132df")
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 2, 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 sys

# Simple profit matrix - 3x4 matrix of profit values
profit_matrix = [
    [10, 20, 30, 40],
    [15, 25, 35, 45],
    [12, 22, 32, 42]
]

# Inversion function that converts profit to cost
# Using a simple approach: max_value - profit
inversion_function = lambda x: 100 - x

ARGS = {
    "profit_matrix": profit_matrix,
    "inversion_function": inversion_function
}
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 list, ndim 2, 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
ComponentWiseCondSpline Not yet run
unverified (harvested, no recorded run)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("0058b6a47fe9f194")
How it ran

no invocation recorded.

ModularShifts Not yet run
unverified (harvested, no recorded run)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("295b9fa436b63536")
How it ran

no invocation recorded.

Munkres Not yet run
unverified (harvested, no recorded run)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("490d0594e6f6579c")
How it ran

no invocation recorded.

_construct_nn Not yet run
unverified (harvested, no recorded run)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("e09f312815d1ee82")
How it ran

no invocation recorded.

compute_mcc Not yet run
unverified (harvested, no recorded run)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("0b6abb112159adb6")
How it ran

no invocation recorded.

correlation Not yet run
unverified (harvested, no recorded run)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("228dbb5491c93115")
How it ran

no invocation recorded.

kaiming_init Not yet run
unverified (harvested, no recorded run)
weirayao/tdrl/LiLY/modules/modular.py
code served (permissive licence) · get_code("a18c3595f7425f34")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 17 distinct functions with an identity, 8 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 8 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.

FunctionAlso present inEarliest-datedOthers named
make_cost_matrix
brick of this paper's assembly: helper
8 other papers
2 dated earlier, 6 later · 2 only as an assembly brick
2007.10930 (2020-07)
Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Coding
2206.02416 (2022-06)
Embrace the Gap: VAEs Perform Independent Mechanism Analysis
2310.15709 (2023-10)
Causal Representation Learning Made Identifiable by Grouping of Observational Variables
and 5 more not named here
correlation
brick of this paper's assembly: helper
3 other papers
3 later · 2 only as an assembly brick
2310.18615 (2023-10)
Temporally Disentangled Representation Learning under Unknown Nonstationarity
2401.14535 (2024-01)
CaRiNG: Learning Temporal Causal Representation under Non-Invertible Generation Process
2409.03142 (2024-09)
Causal Temporal Representation Learning with Nonstationary Sparse Transition
kaiming_init
brick of this paper's assembly: helper
3 other papers
1 dated earlier, 2 later · 3 only as an assembly brick
2111.02338 (2021-11)
Drop, Swap, and Generate: A Self-Supervised Approach for Generating Neural Activity
2303.01052 (2023-03)
Demystifying Causal Features on Adversarial Examples and Causal Inoculation for Robust Network by Adversarial Instrumental Variable Regression
2401.14535 (2024-01)
CaRiNG: Learning Temporal Causal Representation under Non-Invertible Generation Process
Munkres
brick of this paper's assembly: helper
2 other papers
2 later · 2 only as an assembly brick
2310.18615 (2023-10)
Temporally Disentangled Representation Learning under Unknown Nonstationarity
2401.14535 (2024-01)
CaRiNG: Learning Temporal Causal Representation under Non-Invertible Generation Process
GroupLinearLayer
brick of this paper's assembly: model
1 other paper
1 later · 1 only as an assembly brick
2401.14535 (2024-01)
CaRiNG: Learning Temporal Causal Representation under Non-Invertible Generation Process
MBDTransitionPrior
brick of this paper's assembly: helper
1 other paper
1 later · 1 only as an assembly brick
2401.14535 (2024-01)
CaRiNG: Learning Temporal Causal Representation under Non-Invertible Generation Process
NLayerLeakyMLP
brick of this paper's assembly: helper
1 other paper
1 later · 1 only as an assembly brick
2401.14535 (2024-01)
CaRiNG: Learning Temporal Causal Representation under Non-Invertible Generation Process
compute_mcc
brick of this paper's assembly: helper
1 other paper
1 later · 1 only as an assembly brick
2401.14535 (2024-01)
CaRiNG: Learning Temporal Causal Representation under Non-Invertible Generation Process

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

Recently in the field of unsupervised representation learning, strong identifiability results for disentanglement of causally-related latent variables have been established by exploiting certain side information, such as class labels, in addition to independence. However, most existing work is constrained by functional form assumptions such as independent sources or further with linear transitions, and distribution assumptions such as stationary, exponential family distribution. It is unknown whether the underlying latent variables and their causal relations are identifiable if they have arbitrary, nonparametric causal influences in between. In this work, we establish the identifiability theories of nonparametric latent causal processes from their nonlinear mixtures under fixed temporal causal influences and analyze how distribution changes can further benefit the disentanglement. We propose TDRL, a principled framework to recover time-delayed latent causal variables and identify their relations from measured sequential data under stationary environments and under different distribution shifts. Specifically, the framework can factorize unknown distribution shifts into transition distribution changes under fixed and time-varying latent causal relations, and under observation changes in observation. Through experiments, we show that time-delayed latent causal influences are reliably identified and that our approach considerably outperforms existing baselines that do not correctly exploit this modular representation of changes. Our code is available at: https://github.com/weirayao/tdrl.

For agents

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

get_harvested_code_for_paper("2210.13647")
get_code_for_paper("2210.13647")
have("2210.13647")

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

curl https://syntology.ai/api/ran/2210.13647.json

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

[![Syntology run record](https://syntology.ai/api/ran/2210.13647.svg)](https://syntology.ai/paper/2210.13647)

Connect an agent — have() is free.