arxiv: 2511.17902 · v3 · submitted 2025-11-22 · 💻 cs.LG · cs.AI· stat.ML

Statistically-Guided Meta-Learning for Cross-Deployment Activity Recognition in Distributed Fiber-Optic Sensing

Yifan He , Haodong Zhang , Qiuheng Song , Lin Lei , Zhenxuan Zeng , Haoyang He , Hongyan Wu This is my paper

Pith reviewed 2026-05-17 06:36 UTC · model grok-4.3

classification 💻 cs.LG cs.AIstat.ML

keywords distributed fiber optic sensingmeta-learningdomain shiftactivity recognitionprototype learningstatistical guidancecross-deployment adaptationdual-domain learning

0 comments

The pith

DUPLE adapts meta-learning prototypes using statistical guidance from time and frequency domains to handle domain shifts in fiber-optic sensing.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces DUPLE to address activity recognition challenges in distributed fiber-optic sensing when moving to new deployment sites that differ in conditions and have scarce or no labels. It builds multi-prototype class representations that draw on complementary cues from both time and frequency domains. A lightweight statistical guidance step estimates how reliable each domain is for a given sample from its raw signal statistics. Query-adaptive aggregation then selects and combines the most suitable prototypes. Experiments on two real-world cross-deployment benchmarks show gains over standard deep learning and meta-learning baselines in accuracy and stability.

Core claim

DUPLE is a prototype-based meta-learning framework that jointly exploits complementary time- and frequency-domain cues to construct multi-prototype class representations, then uses a lightweight statistical guidance mechanism to estimate the reliability of each domain from raw signal statistics and applies query-adaptive aggregation to select and combine the most relevant prototypes for each query sample under cross-deployment domain shift.

What carries the argument

The query-adaptive aggregation strategy that uses statistical estimates of domain reliability from raw signals to select and combine the most relevant multi-prototype representations for each query.

If this is right

Recognition becomes feasible in new DFOS sites without collecting and labeling fresh data.
Performance remains more stable when environmental conditions or sensor placements differ from the source deployment.
Intra-class signal variations are better captured by maintaining multiple prototypes per class instead of a single average.
The overall system requires less manual intervention for adapting to additional deployments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same statistical guidance idea could transfer to other multi-domain sensing problems that face deployment shifts, such as acoustic or vibration monitoring.
Combining the prototype adaptation with unsupervised domain alignment methods might further reduce the label requirement in extreme shift cases.
Real-time latency measurements on actual fiber-optic hardware would test whether the added statistical computations remain practical for continuous operation.

Load-bearing premise

A lightweight statistical guidance mechanism can reliably estimate the reliability of each domain from raw signal statistics and enable effective query-adaptive prototype selection under cross-deployment domain shift.

What would settle it

An ablation experiment that replaces the statistical guidance with random or fixed domain weighting and checks whether accuracy on the two real-world cross-deployment benchmarks drops to or below the levels of the strong baselines.

Figures

Figures reproduced from arXiv: 2511.17902 by Haodong Zhang, Haoyang He, Hongyan Wu, Lin Lei, Qiuheng Song, Yifan He, Zhenxuan Zeng.

**Figure 2.** Figure 2: Overview of the proposed Dual-Domain Meta-Learning framework. The architecture first integrates temporal and spectral information via a Feature [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Figure 3: A schematic diagram of the Statistical Guidance Network (SGN) is [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗

**Figure 4.** Figure 4: A schematic diagram of the Collaborative Decision module. The time [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗

**Figure 5.** Figure 5: Sensitivity analysis of model performance with respect to support set size [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗

**Figure 6.** Figure 6: t-SNE visualization of the feature embeddings learned by DUPLE (5-shot) on representative held-out deployments. [PITH_FULL_IMAGE:figures/full_fig_p011_6.png] view at source ↗

**Figure 7.** Figure 7: Per-class performance analysis on OSDG1 and OSDG2. (a)-(d) Precision and F1-scores. DUPLE is compared against representative baselines: [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗

read the original abstract

Distributed Fiber Optic Sensing (DFOS) is promising for long-range perimeter security, yet practical deployment faces three key obstacles: severe cross-deployment domain shift, scarce or unavailable labels at new sites, and limited within-class coverage even in source deployments. We propose DUPLE, a prototype-based meta-learning framework tailored for cross-deployment DFOS recognition. The core idea is to jointly exploit complementary time- and frequency-domain cues and adapt class representations to sample-specific statistics: (i) a dual-domain learner constructs multi-prototype class representations to cover intra-class heterogeneity; (ii) a lightweight statistical guidance mechanism estimates the reliability of each domain from raw signal statistics; and (iii) a query-adaptive aggregation strategy selects and combines the most relevant prototypes for each query. Extensive experiments on two real-world cross-deployment benchmarks demonstrate consistent improvements over strong deep learning and meta-learning baselines, achieving more accurate and stable recognition under label-scarce target deployments.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

DUPLE applies dual-domain prototypes and statistical guidance to meta-learning for cross-deployment DFOS, with benchmark gains that may not isolate the guidance contribution under complex shifts.

read the letter

The main takeaway is that this paper adapts prototype-based meta-learning to distributed fiber optic sensing by building multi-prototype class representations from both time and frequency domains, then using raw signal statistics to guide query-adaptive selection and aggregation. It reports consistent improvements on two real-world cross-deployment benchmarks over standard deep learning and meta-learning baselines. That setup directly targets the practical barriers of label scarcity and domain shift when moving DFOS systems to new sites for security monitoring. The dual-domain choice is reasonable because vibration signals carry complementary information across domains, and the multi-prototype design helps cover intra-class variation that single prototypes would miss. Testing on actual deployment-shift data rather than synthetic shifts is a plus for relevance. The framework stays grounded in existing meta-learning ideas without obvious circularity or invented components. The soft spot sits with the statistical guidance step. The claim is that lightweight statistics from the raw signals can estimate per-domain reliability well enough to drive effective adaptation. DFOS shifts often arise from non-stationary environmental factors, installation differences, or higher-order noise structures that basic moments and spectral summaries may not capture reliably. If the guidance mis-ranks domains on many queries, the adaptive aggregation adds little beyond what dual-domain prototypes already provide, and the reported gains would not generalize as cleanly. The abstract and experiments do not yet include enough detail on run counts, statistical tests, or ablations that separate the guidance from the rest of the architecture, so the evidence for that specific mechanism stays thinner than the overall results. This work is aimed at applied researchers and engineers working on sensing systems, domain adaptation for physical signals, or few-shot recognition in real deployments. Readers focused on practical ML for perimeter security or vibration monitoring would find the setup and benchmark results worth examining. It shows honest engagement with the literature on meta-learning and domain shift, with no load-bearing contradictions in the described approach. It deserves a serious referee because the problem matters and the experiments use real data, even if revisions should strengthen the analysis of whether the guidance is the operative factor.

Referee Report

2 major / 1 minor

Summary. The manuscript introduces DUPLE, a prototype-based meta-learning framework for cross-deployment activity recognition in distributed fiber-optic sensing (DFOS). It jointly exploits time- and frequency-domain cues via a dual-domain learner that constructs multi-prototype class representations, employs a lightweight statistical guidance mechanism to estimate per-domain reliability from raw signal statistics, and uses query-adaptive aggregation to select and combine relevant prototypes. Experiments on two real-world cross-deployment benchmarks report consistent improvements over deep learning and meta-learning baselines in label-scarce target deployments.

Significance. If validated, the approach could meaningfully advance practical DFOS applications in perimeter security by mitigating severe domain shift and label scarcity through adaptive, statistics-driven prototype aggregation. The targeted use of complementary domains with statistical guidance offers a focused contribution that may extend to other sensor modalities with non-stationary shifts.

major comments (2)

[Method (statistical guidance component)] The central claim relies on the statistical guidance mechanism (described in the abstract and method) accurately estimating domain reliability from raw signal statistics to enable effective query-adaptive prototype selection. However, DFOS domain shifts frequently arise from non-stationary factors, installation differences, or noise profiles that affect higher-order temporal structure rather than first- or second-order moments; the manuscript provides no analysis, correlation study, or ablation isolating whether these simple statistics reliably predict per-query domain discriminability or whether dual-domain prototypes alone would suffice.
[Experiments] The abstract reports 'consistent improvements' and 'more accurate and stable recognition' on two benchmarks, yet the experimental section supplies no details on the number of runs, standard deviations, statistical significance tests, or controls for potential confounds such as hyperparameter tuning or data partitioning. This absence undermines assessment of whether the gains are robust and attributable to the proposed components rather than experimental variability.

minor comments (1)

[Abstract] The abstract refers to 'two real-world cross-deployment benchmarks' without naming them or briefly characterizing their shift characteristics (e.g., environmental vs. installation differences), which would aid readers in evaluating generalizability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which help improve the clarity and rigor of our work on DUPLE for cross-deployment DFOS activity recognition. We address each major comment below and will revise the manuscript accordingly.

read point-by-point responses

Referee: [Method (statistical guidance component)] The central claim relies on the statistical guidance mechanism (described in the abstract and method) accurately estimating domain reliability from raw signal statistics to enable effective query-adaptive prototype selection. However, DFOS domain shifts frequently arise from non-stationary factors, installation differences, or noise profiles that affect higher-order temporal structure rather than first- or second-order moments; the manuscript provides no analysis, correlation study, or ablation isolating whether these simple statistics reliably predict per-query domain discriminability or whether dual-domain prototypes alone would suffice.

Authors: We appreciate the referee's observation regarding the potential limitations of first- and second-order statistics in capturing non-stationary shifts. The statistical guidance mechanism was designed as a lightweight module to provide sample-specific domain reliability estimates from readily computable raw signal statistics, complementing the dual-domain prototypes. While our empirical results on the two benchmarks indicate that this guidance improves query-adaptive aggregation, we acknowledge the absence of a dedicated correlation or ablation analysis in the current manuscript. In the revision, we will add an ablation comparing the full model against a variant that relies solely on dual-domain prototypes (without statistical guidance) and include a correlation study between the estimated reliabilities and per-query domain accuracies to better substantiate the component's contribution. revision: yes
Referee: [Experiments] The abstract reports 'consistent improvements' and 'more accurate and stable recognition' on two benchmarks, yet the experimental section supplies no details on the number of runs, standard deviations, statistical significance tests, or controls for potential confounds such as hyperparameter tuning or data partitioning. This absence undermines assessment of whether the gains are robust and attributable to the proposed components rather than experimental variability.

Authors: We agree that detailed reporting on experimental variability and controls is necessary to support the claims of consistent improvements. The current manuscript indeed omits these specifics. In the revised version, we will expand the experimental section to report results averaged over multiple independent runs (with the exact number and random seeds specified), include standard deviations, conduct and report statistical significance tests (such as paired t-tests against baselines with p-values), and provide explicit details on the hyperparameter tuning procedure and data partitioning strategy for the cross-deployment benchmarks. These additions will allow better assessment of robustness and attribution to the proposed components. revision: yes

Circularity Check

0 steps flagged

No significant circularity; framework components are independently motivated and non-reductive

full rationale

The paper introduces DUPLE as a composite meta-learning framework whose three stated elements (dual-domain multi-prototype learner, statistical guidance from raw-signal moments, and query-adaptive prototype aggregation) are described as distinct, complementary mechanisms motivated by the specific challenges of DFOS cross-deployment shift. No equations, definitions, or self-citations are shown that would make any component equivalent to its own inputs by construction, nor is any fitted parameter relabeled as a prediction. The abstract and high-level description present the statistical guidance as an additional lightweight module rather than a tautological re-expression of the prototypes themselves. Because the derivation chain therefore remains additive and externally benchmarked rather than self-referential, the work is self-contained against the listed circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The abstract does not specify any free parameters, axioms, or new entities. The framework relies on standard meta-learning concepts but details are not provided.

pith-pipeline@v0.9.0 · 5483 in / 1214 out tokens · 80033 ms · 2026-05-17T06:36:41.543226+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

a lightweight statistical guidance mechanism estimates the reliability of each domain from raw signal statistics
IndisputableMonolith/Foundation/ArithmeticFromLogic.lean embed_strictMono_of_one_lt unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

dual-domain learner constructs multi-prototype class representations... query-adaptive aggregation strategy

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages · 1 internal anchor

[1]

, author Dou, S

author Ajo-Franklin, J.B. , author Dou, S. , author Lindsey, N.J. , author Monga, I. , author Tracy, C. , author Robertson, M. , author Rodriguez Tribaldos, V. , author Ulrich, C. , author Freifeld, B. , author Daley, T. , et al., year 2019 . title Distributed acoustic sensing using dark fiber for near-surface characterization and broadband seismic event ...

work page 2019
[2]

, author Shelhamer, E

author Allen, K. , author Shelhamer, E. , author Shin, H. , author Tenenbaum, J. , year 2019 . title Infinite mixture prototypes for few-shot learning , in: booktitle International conference on machine learning , organization PMLR . pp. pages 232--241

work page 2019
[3]

Meta-learning with differentiable closed-form solvers

author Bertinetto, L. , author Henriques, J.F. , author Torr, P.H. , author Vedaldi, A. , year 2018 . title Meta-learning with differentiable closed-form solvers . journal arXiv preprint arXiv:1805.08136

work page internal anchor Pith review Pith/arXiv arXiv 2018
[4]

, author Guestrin, C

author Chen, T. , author Guestrin, C. , year 2016 . title Xgboost: A scalable tree boosting system , in: booktitle Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining , pp. pages 785--794

work page 2016
[5]

, author Abbeel, P

author Finn, C. , author Abbeel, P. , author Levine, S. , year 2017 . title Model-agnostic meta-learning for fast adaptation of deep networks , in: booktitle International conference on machine learning , organization PMLR . pp. pages 1126--1135

work page 2017
[6]

, year 2017

author Hartog, A.H. , year 2017 . title An introduction to distributed optical fibre sensors . publisher CRC press

work page 2017
[7]

, author Liu, Q

author He, Z. , author Liu, Q. , year 2021 . title Optical fiber distributed acoustic sensors: A review . journal Journal of Lightwave Technology volume 39 , pages 3671--3686

work page 2021
[8]

, author Liang, S

author Jia, H. , author Liang, S. , author Lou, S. , author Sheng, X. , year 2019 a. title A k -nearest neighbor algorithm-based near category support vector machine method for event identification of -otdr . journal IEEE Sensors Journal volume 19 , pages 3683--3689

work page 2019
[9]

, author Lou, S

author Jia, H. , author Lou, S. , author Liang, S. , author Sheng, X. , year 2019 b. title Event identification by f-elm model for - OTDR fiber-optic distributed disturbance sensor . journal IEEE Sensors Journal volume 20 , pages 1297--1305

work page 2019
[10]

, author Wang, L

author Khaki, S. , author Wang, L. , author Archontoulis, S.V. , year 2020 . title A cnn-rnn framework for crop yield prediction . journal Frontiers in Plant Science volume 10 , pages 1750

work page 2020
[11]

, author Ar k, S

author Lim, B. , author Ar k, S. \"O . , author Loeff, N. , author Pfister, T. , year 2021 . title Temporal fusion transformers for interpretable multi-horizon time series forecasting . journal International journal of forecasting volume 37 , pages 1748--1764

work page 2021
[12]

, author Dawe, T.C

author Lindsey, N.J. , author Dawe, T.C. , author Ajo-Franklin, J.B. , year 2019 . title Illuminating seafloor faults and ocean dynamics with dark fiber distributed acoustic sensing . journal Science volume 366 , pages 1103--1107

work page 2019
[13]

, author Deligiannis, N

author Luong, H.V. , author Deligiannis, N. , author Wilhelm, R. , author Drapp, B. , year 2023 . title Few-shot classification with meta-learning for urban infrastructure monitoring using distributed acoustic sensing . journal Sensors volume 24 , pages 49

work page 2023
[14]

, author Bae, T

author Madsen, C. , author Bae, T. , author Snider, T. , year 2007 . title Intruder signature analysis from a phase-sensitive distributed fiber-optic perimeter sensor , in: booktitle Fiber Optic Sensors and Applications V , organization SPIE . pp. pages 141--148

work page 2007
[15]

, author Jain, L

author Medsker, L.R. , author Jain, L. , et al., year 2001 . title Recurrent neural networks . journal Design and applications volume 5 , pages 2

work page 2001
[16]

, author Soto, M.A

author Mu \ n oz, F. , author Soto, M.A. , year 2022 . title Enhancing fibre-optic distributed acoustic sensing capabilities with blind near-field array signal processing . journal Nature Communications volume 13 , pages 4019

work page 2022
[17]

, author Feng, H

author Qu, Z. , author Feng, H. , author Zeng, Z. , author Zhuge, J. , author Jin, S. , year 2010 . title A svm-based pipeline leakage detection and pre-warning system . journal Measurement volume 43 , pages 513--519

work page 2010
[18]

, author Raghu, M

author Raghu, A. , author Raghu, M. , author Bengio, S. , author Vinyals, O. , year 2019 . title Rapid learning or feature reuse? towards understanding the effectiveness of maml . journal arXiv preprint arXiv:1909.09157

work page arXiv 2019
[19]

, author Wang, Z

author Rao, Y. , author Wang, Z. , author Wu, H. , author Ran, Z. , author Han, B. , year 2021 . title Recent advances in phase-sensitive optical time domain reflectometry ( - OTDR ) . journal photonic sensors volume 11 , pages 1--30

work page 2021
[20]

, year 2020

author Sherstinsky, A. , year 2020 . title Fundamentals of recurrent neural network (rnn) and long short-term memory (lstm) network . journal Physica D: Nonlinear Phenomena volume 404 , pages 132306

work page 2020
[21]

, author Swersky, K

author Snell, J. , author Swersky, K. , author Zemel, R. , year 2017 . title Prototypical networks for few-shot learning . journal Advances in neural information processing systems volume 30

work page 2017
[22]

, author Macias-Guarasa, J

author Tejedor, J. , author Macias-Guarasa, J. , author Martins, H.F. , author Martin-Lopez, S. , author Gonzalez-Herraez, M. , year 2019 . title A contextual gmm-hmm smart fiber optic surveillance system for pipeline integrity threat detection . journal Journal of Lightwave Technology volume 37 , pages 4514--4522

work page 2019
[23]

, author Macias-Guarasa, J

author Tejedor, J. , author Macias-Guarasa, J. , author Martins, H.F. , author Piote, D. , author Pastor-Graells, J. , author Martin-Lopez, S. , author Corredera, P. , author Gonzalez-Herraez, M. , year 2017 . title A novel fiber optic based surveillance system for prevention of pipeline integrity threats . journal Sensors volume 17 , pages 355

work page 2017
[24]

, author Kazei, V

author Titov, A. , author Kazei, V. , author AlDawood, A. , author Alfataierge, E. , author Bakulin, A. , author Osypov, K. , year 2022 . title Quantification of das vsp quality: Snr vs. log-based metrics . journal Sensors volume 22 , pages 1027

work page 2022
[25]

, author Shazeer, N

author Vaswani, A. , author Shazeer, N. , author Parmar, N. , author Uszkoreit, J. , author Jones, L. , author Gomez, A.N. , author Kaiser, . , author Polosukhin, I. , year 2017 . title Attention is all you need . journal Advances in neural information processing systems volume 30

work page 2017
[26]

, author Blundell, C

author Vinyals, O. , author Blundell, C. , author Lillicrap, T. , author Wierstra, D. , et al., year 2016 . title Matching networks for one shot learning . journal Advances in neural information processing systems volume 29

work page 2016
[27]

, author Yang, Y

author Wang, J. , author Yang, Y. , author Mao, J. , author Huang, Z. , author Huang, C. , author Xu, W. , year 2016 . title Cnn-rnn: A unified framework for multi-label image classification , in: booktitle Proceedings of the IEEE conference on computer vision and pattern recognition , pp. pages 2285--2294

work page 2016
[28]

, author Lu, B

author Wang, Z. , author Lu, B. , author Ye, Q. , author Cai, H. , year 2020 . title Recent progress in distributed fiber acoustic sensing with - OTDR . journal Sensors volume 20 , pages 6594

work page 2020
[29]

, author Chen, J

author Wu, H. , author Chen, J. , author Liu, X. , author Xiao, Y. , author Wang, M. , author Zheng, Y. , author Rao, Y. , year 2019 a. title One-dimensional cnn-based intelligent recognition of vibrations in pipeline monitoring with das . journal Journal of Lightwave Technology volume 37 , pages 4359--4366

work page 2019
[30]

, author Gan, D

author Wu, H. , author Gan, D. , author Xu, C. , author Liu, Y. , author Liu, X. , author Song, Y. , author Rao, Y. , year 2022 . title Improved generalization in signal identification with unsupervised spiking neuron networks for fiber-optic distributed acoustic sensor . journal Journal of Lightwave Technology volume 40 , pages 3072--3083

work page 2022
[31]

, author Liu, X

author Wu, H. , author Liu, X. , author Xiao, Y. , author Rao, Y. , year 2019 b. title A dynamic time sequence recognition and knowledge mining method based on the hidden markov models (hmms) for pipeline safety monitoring with -otdr . journal Journal of Lightwave Technology volume 37 , pages 4991--5000

work page 2019
[32]

, author Zhou, B

author Wu, H. , author Zhou, B. , author Zhu, K. , author Shang, C. , author Tam, H.Y. , author Lu, C. , year 2021 . title Pattern recognition in distributed fiber-optic acoustic sensor using an intensity and phase stacked convolutional neural network with data augmentation . journal Optics express volume 29 , pages 3269--3283

work page 2021
[33]

, author Guan, J

author Xu, C. , author Guan, J. , author Bao, M. , author Lu, J. , author Ye, W. , year 2017 . title Pattern recognition based on enhanced multifeature parameters for vibration events in - OTDR distributed optical fiber sensing system . journal Microwave and Optical Technology Letters volume 59 , pages 3134--3141

work page 2017
[34]

, author Guan, J

author Xu, C. , author Guan, J. , author Bao, M. , author Lu, J. , author Ye, W. , year 2018 . title Pattern recognition based on time-frequency analysis and convolutional neural networks for vibrational events in -otdr . journal Optical Engineering volume 57 , pages 016103--016103

work page 2018
[35]

, author Hu, H

author Ye, H.J. , author Hu, H. , author Zhan, D.C. , author Sha, F. , year 2020 . title Few-shot learning via embedding adaptation with set-to-set functions , in: booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. pages 8808--8817

work page 2020
[36]

, author Gao, J

author Zhang, H. , author Gao, J. , author Hong, B. , year 2022 . title -otdr signal identification method based on multimodal fusion . journal Sensors volume 22 , pages 8795

work page 2022
[37]

, author Chiang, C.Y

author Zhong, R. , author Chiang, C.Y. , author Jaber, M. , author De Wilde, R. , author Hayward, P. , year 2025 . title Intelligent travel activity monitoring: Generalized distributed acoustic sensing approaches . journal arXiv preprint arXiv:2506.10237

work page arXiv 2025
[38]

, author Zhang, S

author Zhou, H. , author Zhang, S. , author Peng, J. , author Zhang, S. , author Li, J. , author Xiong, H. , author Zhang, W. , year 2021 . title Informer: Beyond efficient transformer for long sequence time-series forecasting , in: booktitle Proceedings of the AAAI conference on artificial intelligence , pp. pages 11106--11115

work page 2021