PONE-D-25-38730KALFormer: Knowledge-Augmented Attention Learning for Long-Term Time Series Forecasting with TransformerPLOS ONE

Dear Dr. Zhang,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process. Both reviewers raised important points: I strongly recommend following the suggestions of Reviewer #1, particularly the ones about technical issues (equations, methodology, analysis of results), but also consider answering the more general comments of Reviewer #2 or providing counterarguments where appliable. Also please review your paper for small typos and mistakes, e.g: addresses of authors sometimes have spaces before/after the ZIP code, sometimes not. Some references' titles contains abbreviations and acronyms: should those be capitalized? E.g. reference 21 (A3t-gcn), reference 12 (Multi-head attention). Also, "Recurrent Neural Network" is capitalized in reference 25 but not in others, e.g. 30. Just to be safe, please consider using a reference management software.  Please also see the reviewers' comments on the quality of the figures, For example, figure 3 is blurry and the numbers on the table are tiny.

Please submit your revised manuscript by Oct 21 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Rafael Duarte Coelho dos Santos, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match.

When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

3. Thank you for stating the following financial disclosure:

National Natural Science Foundation of China�GuizhouUniversity of Traditional Chinese MMedicine

Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

If this statement is not correct you must amend it as needed.

Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf.

4. Thank you for uploading your study's underlying data set. Unfortunately, the repository you have noted in your Data Availability statement does not qualify as an acceptable data repository according to PLOS's standards.

At this time, please upload the minimal data set necessary to replicate your study's findings to a stable, public repository (such as figshare or Dryad) and provide us with the relevant URLs, DOIs, or accession numbers that may be used to access these data. For a list of recommended repositories and additional information on PLOS standards for data deposition, please see https://journals.plos.org/plosone/s/recommended-repositories.

5. We note you have included a table to which you do not refer in the text of your manuscript. Please ensure that you refer to Table 3 in your text; if accepted, production will need this reference to link the reader to the Table.

6. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise. 

Additional Editor Comments:

Reviewer #1:

Please verify the detailed suggestions on technical aspects of the paper, particularly formulation and statistical analysis, and also make sure the public code is the one related to the paper and fully documented.

Reviewer #2:

Please check and address the comments, in particular comment 1 (related to the whole approach in general -- may be necessary to compare with the current state of the art in time series forecasting) and comment 5 (more details on the interpretation pf the results).

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The manuscript presents KALFormer, a hybrid model integrating LSTM, multi-head attention, knowledge-augmented networks (KAN), and Transformer modules for long-term time series forecasting. The idea of combining sequential modeling with external knowledge augmentation is novel and potentially valuable. However, in its current form the paper has several important issues that limit reproducibility, clarity, and rigor. Below I outline my detailed comments.

1. Technical soundness

The overall framework is reasonable and produces encouraging results across multiple benchmark datasets (Traffic, Electricity, Weather, ETTh1/2, ETTm1/2).

However, there are issues with the mathematical formulation. Equation (3) misrepresents the standard scaled dot-product attention. The correct formulation is:

Attention(,,)=softmax(⊤)Attention(Q,K,V)=softmax(d ​QK⊤ ​)V

The manuscript currently places the softmax outside the multiplication with V, which is not correct. This should be revised to avoid confusion and ensure technical accuracy.

The KAN component is under-specified. It is not clear how the knowledge graph is constructed for each dataset:

What constitutes nodes and edges?

How are edge weights computed (correlation, distance, domain rules)?

Are graphs dataset-specific or universal?

How many GCN layers are used and how is the fusion weight initialized and trained?

Without these details, the contribution of the “knowledge” component cannot be independently validated or reproduced.

2. Statistical analysis

The manuscript reports MSE and MAE, which are standard metrics for regression tasks. However, there are inconsistencies between results across different tables. For example, the ETTh1 MSE in Table 3 differs significantly from the value reported in Table 5. Please clarify whether these correspond to different prediction horizons, different data splits, or a reporting error.

In Table 4 (ablation study), an “Accuracy (%)” metric is introduced. Accuracy is not standard for time series regression, and no definition is provided. Does this refer to directional accuracy (predicting the correct trend) or tolerance-based accuracy (predictions within a margin)? This must be defined formally in the Methods section. If it cannot be rigorously justified, it should be replaced with standard error metrics (RMSE, MAPE, or directional metrics explicitly stated).

There is no discussion of statistical significance or confidence intervals. To strengthen the claims of superiority over baselines, the authors should consider including repeated runs with error bars or tests such as the Diebold–Mariano test.

3. Data and code availability

The code is shared via GitHub, which is commendable. However, the manuscript describes KALFormer but the linked repository is titled FusionLSTM_KT. It is unclear whether this repository contains the exact version of KALFormer used to generate the reported results. To satisfy reproducibility requirements, please ensure the repository contains:

A clearly named KALFormer implementation.

Configuration files and scripts for each dataset/horizon.

Instructions to reproduce Tables 3–5 exactly.

The Data Availability Statement does not meet PLOS ONE policy. The Kaggle link contains a misspelled username (“anonuymous”) and has no persistent DOI. PLOS requires that all underlying data be deposited in a stable, publicly accessible repository (e.g., Zenodo, OSF, Figshare, or Kaggle with DOI). Preprocessing scripts should also be provided so other researchers can replicate the train/validation/test splits and normalization steps.

4. Experimental setup and reproducibility

The manuscript notes that datasets are split 8:1:1, but does not specify if the split is time-ordered (which is typical for forecasting tasks) or random. This affects the validity of the evaluation.

Details such as input sequence length (look-back window), output horizons (96, 192, etc.), and normalization (per-feature vs per-series, fit on training only or across the whole dataset) should be reported.

Missing value handling is not discussed. Were missing data imputed, and if so, how?

Hyperparameters such as learning rate, dropout, number of epochs, batch size, optimizer, and weight decay should be provided. Currently, only the hardware setup is described.

Providing a summary table of these details would substantially improve reproducibility.

5. Ablation study and interpretation

The ablation study shows that KAN + Attention (without Transformer) reduces performance, but the full model improves once the Transformer is added. This is attributed to “feature inconsistency,” but this explanation is speculative without additional baselines.

A stronger comparison would include:

A “KAN only” baseline.

A “KAN + simple concatenation + MLP” baseline.

This would clarify whether the Transformer specifically enables effective integration of KAN features.

The narrative should also reconcile why KAN appears to degrade performance in some ablations but still improves the full model.

6. Presentation and writing

The manuscript is generally intelligible and written in clear English. However, minor issues remain:

Typos such as “anonuymous” and “MMedicine.”

Inconsistent capitalization of terms (KAN, Transformer, MHA).

Figures should be provided as high-resolution images per journal guidelines.

Correcting these will improve readability.

7. Funding and policy compliance

The funding statement currently lists organizations but does not provide all details required by PLOS ONE (initials of funded authors, grant numbers, funder URLs, and explicit statement of funders’ role). Please expand this section accordingly.

Summary:

This manuscript proposes a creative hybrid model for long-term time series forecasting and shows promising empirical results. However, major revisions are required before the manuscript can be considered for publication. The main issues involve data and code availability, mathematical correctness, statistical rigor, and insufficient methodological detail for reproducibility. Addressing these points will significantly strengthen the work and ensure compliance with PLOS ONE standards.

Reviewer #2: The authors have focused on time series forecasting, an area that has already seen extensive research and progress.

The reviewer has the following concerns:

1. The problem described in the paper is inherently a multivariate time series forecasting task, for which significant research already exists. However, starting from the introduction, the authors frame the problem as forecasting a target series while treating other time series merely as influencing factors. This approach raises questions regarding the fairness of the problem analysis and validation.

2. In the validation of the models, both the proposed model and the baseline comparisons lack details regarding model architecture parameters and key configuration settings. As a result, the fairness of the results cannot be guaranteed.

3. The use of LSTM for feature extraction from the time series before applying the Transformer may seem reasonable. However, the reviewers question whether this modification is truly beneficial—given that the Transformer itself is capable of extracting temporal features and its strengths lie in self-attention and cross-attention mechanisms. From the perspectives of performance and effectiveness, is such a design choice justified?

4. The connections between modules in the proposed model need further refinement, particularly in terms of the input and output features.

5. The analysis and discussion of the experimental findings are overly brief, especially in explaining the reasons behind the performance improvement. More in-depth interpretation is needed.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

KALFormer: Knowledge-Augmented Attention Learning for Long-Term Time Series Forecasting with Transformer

PONE-D-25-38730R1

Dear Dr. Zhang,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Rafael Duarte Coelho dos Santos, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

The corrections and improvements are satisfactory.

Reviewers' comments: