SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2109.11428 · 2021

An Evaluation of Anomaly Detection and Diagnosis in Multivariate Time Series

arXiv · PDF · Open in the Atlas

Code that ran

7 of 7 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; 1 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 1 failure of our instrument; 0 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/2109.11428.json · badge: /api/ran/2109.11428.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
astha-chem/mvts-ano-eval canonical (classifier label)
classifier basis: census
7 ran · 0 no recorded run
FunctionStatusWhere it lives
adjust_predicts Ran
ran on a synthesized input
astha-chem/mvts-ano-eval/src/algorithms/omni_anomaly/eval_methods.py
code served (permissive licence) · get_code("e9d1085c7f3570bf")
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 ndarray, ndim 1, dtype family float, shape (8,) 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 numpy as np

# Create sample data for the function
score = np.array([0.1, 0.5, 0.3, 0.8, 0.2, 0.9, 0.4, 0.1], dtype=np.float32)
label = np.array([0.0, 0.0, 1.0, 1.0, 0.0, 0.0, 1.0, 1.0], dtype=np.float32)
threshold = 0.5
pred = None
calc_latency = False
point_adjust = True

ARGS = {
    "score": score,
    "label": label,
    "threshold": threshold,
    "pred": pred,
    "calc_latency": calc_latency,
    "point_adjust": point_adjust
}
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 ndarray, ndim 1, dtype family float, shape (8,) — shape (8,) 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
calc_point2point Ran
ran, honoured the contract we drafted
astha-chem/mvts-ano-eval/src/algorithms/omni_anomaly/eval_methods.py
code served (permissive licence) · get_code("dc9eba4733cf0ef2")
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 numpy as np

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

# Create sample binary prediction and actual arrays
# These should be binary (0 or 1) values since the function treats them as classification labels
predict = np.array([1, 0, 1, 1, 0, 1, 0, 0], dtype=np.int32)
actual = np.array([1, 0, 1, 0, 0, 1, 1, 0], dtype=np.int32)

ARGS = {
    "predict": predict,
    "actual": actual
}
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, 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
calc_seq Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
astha-chem/mvts-ano-eval/src/algorithms/omni_anomaly/eval_methods.py
code served (permissive licence) · get_code("ffb9d8932a2a0060")
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 numpy as np

# Create sample data for calc_seq function
np.random.seed(0)

# Generate synthetic anomaly scores (continuous values)
score = np.random.rand(20) * 2  # Random scores between 0 and 2

# Generate synthetic labels (binary: 0 for normal, 1 for anomaly)
label = np.zeros(20)
label[5:8] = 1  # Create an anomaly segment
label[15:18] = 1  # Create another anomaly segment

# Set threshold for anomaly detection
threshold = 0.5

ARGS = {
    "score": score,
    "label": label,
    "threshold": threshold,
    "calc_latency": False
}
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, 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
get_best_config Ran
ran on a synthesized input
astha-chem/mvts-ano-eval/configs.py
code served (permissive licence) · get_code("5cff0c62ca6be16c")
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 dict. 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)

ARGS = {
    'algo_name': 'AutoEncoder_recon_all',
    'ds_name': 'swat'
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
get_sub_seqs Ran
ran on a synthesized input
astha-chem/mvts-ano-eval/src/algorithms/algorithm_utils.py
code served (permissive licence) · get_code("370f5ea00f912a03")
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 ndarray, ndim 3, dtype family float, shape (4, 4, 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 numpy as np

# Construct x_arr: a time-series array with shape (time, channels)
# Using shape (10, 3) - 10 time points, 3 channels
x_arr = np.random.randn(10, 3).astype(np.float32)

# Define seq_len: the window size for subsequences
seq_len = 4

# Define stride: how many time points to move between subsequences
stride = 2

# Define start_discont: starting points of discontinuities
# Using empty array as default, but can also test with values
start_discont = np.array([])

# Alternative with discontinuities (optional, but shown as possibility):
# start_discont = np.array([6, 8])

ARGS = {
    "x_arr": x_arr,
    "seq_len": seq_len,
    "stride": stride,
    "start_discont": start_discont,
}
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 ndarray, ndim 3, dtype family float, shape (4, 4, 3) — shape (4, 4, 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
get_thres_config Ran
ran on a synthesized input
astha-chem/mvts-ano-eval/configs.py
code served (permissive licence) · get_code("cd2efafd2f0a09ba")
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 dict. 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 copy

n_dims = {'swat-long': 51, 'swat': 51, 'wadi': 123, 'damadics': 32, 'damadics-s': 32, 'msl': 55, 'smap': 25, 'smd': 38, 'skab': 8}

seq_lens = {'swat-long': 100, 'swat': 100, 'wadi': 30, 'damadics': 100, 'damadics-s': 100, 'msl': 100, 'smap': 100, 'smd': 100, 'skab': 100}

long_windows = {'swat-long': 100000, 'swat': 100000, 'wadi': 100000, 'damadics': 100000, 'msl': 2000, 'smap': 2000, 'smd': 25000, 'damadics-s': 100000, 'skab': 100}

kernel_sigmas = {'swat-long': 120, 'swat': 120, 'wadi': 120, 'damadics': 5, 'msl': 10, 'smap': 10, 'smd': 1, 'damadics-s': 5, 'skab': 1}

default_thres_config = {'top_k_time': {}, 'best_f1_test': {'exact_pt_adj': True}, 'thresholded_score': {}, 'tail_prob': {'tail_prob': 2}, 'tail_prob_1': {'tail_prob': 1}, 'tail_prob_2': {'tail_prob': 2}, 'tail_prob_3': {'tail_prob': 3}, 'tail_prob_4': {'tail_prob': 4}, 'tail_prob_5': {'tail_prob': 5}, 'dyn_gauss': {'long_window': 10000, 'short_window': 1, 'kernel_sigma': 10}, 'nasa_npt': {'batch_size': 70, 'window_size': 30, 'telem_only': True, 'smoothing_perc': 0.005, 'l_s': 250, 'error_buffer': 5, 'p': 0.05}}

def get_thres_config(ds_name):
    thres_config = copy.deepcopy(default_thres_config)
    thres_config['dyn_gauss']['short_window'] = 1
    thres_config['dyn_gauss']['long_window'] = long_windows[ds_name]
    thres_config['dyn_gauss']['kernel_sigma'] = kernel_sigmas[ds_name]
    for i in range(1, 6):
        if ds_name not in ['msl', 'smap']:
            thres_config['tail_prob_' + str(i)]['tail_prob'] = i * n_dims[ds_name]
        else:
            thres_config['tail_prob_' + str(i)]['tail_prob'] = i
    thres_config['nasa_npt']['l_s'] = seq_lens[ds_name]
    return thres_config

ARGS = {
    'ds_name': 'swat'
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
get_train_data_loaders Ran
ran on a synthesized input
astha-chem/mvts-ano-eval/src/algorithms/algorithm_utils.py
code served (permissive licence) · get_code("9e8f9ac953549d35")
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 numpy as np
import torch
from torch.utils.data import DataLoader

torch.manual_seed(0)

# Create sample sequence data
# Shape: (num_samples, sequence_length) where each row is a sample/sequence
x_seqs = np.random.randn(10, 4).astype(np.float32)

# Define arguments
ARGS = {
    "x_seqs": x_seqs,
    "batch_size": 2,
    "splits": [0.6, 0.2, 0.2],
    "seed": 42,
    "shuffle": True,
    "usetorch": 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, 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

Also present in other papers' code

Of 7 samples lifted from this paper's linked repositories, 7 carry a function identity and 0 do not. Of the 7 distinct functions with an identity, 3 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 3 later) and 4 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
calc_point2point 3 other papers
2 dated earlier, 1 later
2009.02040 (2020-09)
Multivariate Time-series Anomaly Detection via Graph Attention Network
2108.06816 (2021-08)
Weakly Supervised Temporal Anomaly Segmentation with Dynamic Time Warping
2605.14632 (2026-05)
DRL-STAF: A Deep Reinforcement Learning Framework for State-Aware Forecasting of Complex Multivariate Hidden Markov Processes
calc_seq 2 other papers
1 dated earlier, 1 later
2108.06816 (2021-08)
Weakly Supervised Temporal Anomaly Segmentation with Dynamic Time Warping
2605.14632 (2026-05)
DRL-STAF: A Deep Reinforcement Learning Framework for State-Aware Forecasting of Complex Multivariate Hidden Markov Processes
get_sub_seqs 1 other paper
1 later
2207.12201 (2022-07)
Calibrated One-class Classification for Unsupervised Time Series Anomaly Detection

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

Several techniques for multivariate time series anomaly detection have been proposed recently, but a systematic comparison on a common set of datasets and metrics is lacking. This paper presents a systematic and comprehensive evaluation of unsupervised and semi-supervised deep-learning based methods for anomaly detection and diagnosis on multivariate time series data from cyberphysical systems. Unlike previous works, we vary the model and post-processing of model errors, i.e. the scoring functions independently of each other, through a grid of 10 models and 4 scoring functions, comparing these variants to state of the art methods. In time-series anomaly detection, detecting anomalous events is more important than detecting individual anomalous time-points. Through experiments, we find that the existing evaluation metrics either do not take events into account, or cannot distinguish between a good detector and trivial detectors, such as a random or an all-positive detector. We propose a new metric to overcome these drawbacks, namely, the composite F-score ($Fc_1$), for evaluating time-series anomaly detection. Our study highlights that dynamic scoring functions work much better than static ones for multivariate time series anomaly detection, and the choice of scoring functions often matters more than the choice of the underlying model. We also find that a simple, channel-wise model - the Univariate Fully-Connected Auto-Encoder, with the dynamic Gaussian scoring function emerges as a winning candidate for both anomaly detection and diagnosis, beating state of the art algorithms.

For agents

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

get_harvested_code_for_paper("2109.11428")
get_code_for_paper("2109.11428")
have("2109.11428")

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

curl https://syntology.ai/api/ran/2109.11428.json

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

[![Syntology run record](https://syntology.ai/api/ran/2109.11428.svg)](https://syntology.ai/paper/2109.11428)

Connect an agent — have() is free.