pith. sign in

arxiv: 2605.27044 · v2 · pith:WQCP2W7Inew · submitted 2026-05-26 · 💻 cs.AI

BatteryMFormer: Multi-level Learning for Battery Degradation Trajectory Forecasting

Ruifeng Tan , Jintao Dong , Weixiang Hong , Jia Li , Jiaqiang Huang , Tong-Yi Zhang This is my paper

Pith reviewed 2026-06-29 17:45 UTC · model grok-4.3

classification 💻 cs.AI
keywords battery degradationtrajectory forecastingstate of healthTransformermulti-level learningSOC variationsearly predictionaging conditions
0
0 comments X

The pith

BatteryMFormer forecasts full battery degradation trajectories from early data by explicitly modeling multi-level aging patterns and SOC-localized variations.

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

The paper aims to improve early battery degradation trajectory forecasting, which predicts the full state-of-health path using only initial operational records. Battery data contain regularities shared within specific aging conditions and across batteries, plus voltage-current changes that concentrate in particular state-of-charge intervals. Prior methods do not target these traits directly. BatteryMFormer adds three elements to a Transformer: an aging-condition-aware decoder that supplies condition-specific priors through queries and attention, a memory bank of degradation prototypes that can be retrieved for guidance, and a dual-view encoder that processes both time dynamics and charge-interval features from voltage and current series. Experiments across four battery domains show consistent gains over baselines.

Core claim

BatteryMFormer is a multi-level Transformer for early BDTF that integrates an aging-condition-aware decoder injecting priors via informed queries and attention, a meta degradation pattern memory that learns and retrieves trajectory prototypes, and a dual-view encoder jointly capturing temporal dynamics and SOC-localized variations from voltage-current time series, thereby addressing the multi-level structure and localized profile changes that existing methods overlook.

What carries the argument

BatteryMFormer architecture, which adds aging-condition-informed queries and attention, meta degradation pattern memory retrieval, and dual-view encoding of temporal and SOC-specific signals to a Transformer backbone.

If this is right

  • Early trajectory forecasts become reliable enough to inform manufacturing adjustments and usage optimization before full degradation occurs.
  • Shared patterns across batteries allow the model to generalize when new units enter the same aging condition.
  • Focusing on specific SOC intervals reduces error accumulation in long-horizon forecasts.
  • Prototype retrieval from the meta memory supports consistent predictions even when early data are sparse.
  • The same design yields gains on multiple battery domains, suggesting transferability within the field.

Where Pith is reading between the lines

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

  • The dual-view encoder idea could be tested on other sensor streams where certain operating regimes drive most of the signal change.
  • If the meta memory proves central, replacing it with learned prototypes from physics-based simulations might reduce the need for large labeled datasets.
  • Extending the aging-condition queries to include temperature or load history would be a direct next step if those variables show similar clustering.
  • Real-time deployment logs from fielded packs could serve as a falsification test for whether the learned prototypes remain stable outside laboratory conditions.

Load-bearing premise

The two data characteristics of multi-level structure and SOC-localized variations are the main reasons prior methods fall short, and the three added components directly close that gap.

What would settle it

An ablation study on the four battery domains in which removing the aging-condition-aware decoder, the meta memory, or the dual-view encoder individually brings performance back to the level of the strongest baseline.

Figures

Figures reproduced from arXiv: 2605.27044 by Jia Li, Jiaqiang Huang, Jintao Dong, Ruifeng Tan, Tong-Yi Zhang, Weixiang Hong.

Figure 1
Figure 1. Figure 1: Motivation for multi-level learning in BDTF. (a) An [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: SOC-localized degradation signatures in voltage– [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: An overview of BatteryMFormer. Left: dual-view encoder (temporal and SOC views). Middle: aging-condition-aware [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Performance of the top-three models as the number of usable early cycles increases. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Case study on three representative test batter [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Differential voltage analysis and average attention [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
read the original abstract

Early battery degradation trajectory forecasting (BDTF), which predicts the full-life state-of-health trajectory from early operational data, is critical for battery optimization, manufacturing, and deployment. Battery degradation data exhibit two key characteristics. First, degradation data present a multi-level structure, including regularities shared within aging conditions and trajectory patterns shared across batteries. Second, degradation-related variations in voltage-current profiles are often localized to specific state of charge (SOC) intervals. Existing approaches often fail to explicitly model these characteristics. To bridge this gap, we propose BatteryMFormer, a multi-level Transformer for early BDTF. BatteryMFormer integrates (1) an aging-condition-aware decoder that injects aging-condition priors via aging-condition-informed queries and aging-condition-aware attention, (2) a meta degradation pattern memory that learns and retrieves trajectory prototypes to guide long-horizon forecasting, and (3) a dual-view encoder that jointly captures temporal dynamics and SOC-localized variations from voltage and current time series. Extensive experiments on four battery domains show that BatteryMFormer consistently outperforms state-of-the-art baselines, marking a significant step toward reliable BDTF. Our code is available at https://github.com/Ruifeng-Tan/BatteryMFormer.

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.

Referee Report

2 major / 0 minor

Summary. The paper proposes BatteryMFormer, a multi-level Transformer for early battery degradation trajectory forecasting (BDTF). It identifies two key data characteristics—multi-level structure (regularities within aging conditions and shared trajectory patterns across batteries) and SOC-localized variations in voltage-current profiles—and introduces three components to address them: an aging-condition-aware decoder (using aging-condition-informed queries and attention), a meta degradation pattern memory (for learning and retrieving trajectory prototypes), and a dual-view encoder (capturing temporal dynamics and SOC-localized variations). The central claim is that experiments on four battery domains show consistent outperformance over state-of-the-art baselines, advancing reliable BDTF.

Significance. If the claimed outperformance holds with rigorous validation, the work could advance BDTF by explicitly targeting multi-level and SOC-specific data traits that prior methods overlook, with potential benefits for battery optimization and deployment. Code availability supports reproducibility, which strengthens the contribution if the experiments are detailed and falsifiable.

major comments (2)
  1. [Abstract] Abstract: The central claim that BatteryMFormer 'consistently outperforms state-of-the-art baselines' is presented without any quantitative results, specific baseline names, error metrics, error bars, or validation protocol, making the claim unverifiable from the provided evidence and load-bearing for the significance statement.
  2. [Abstract] Abstract: The manuscript asserts that the two data characteristics (multi-level structure and SOC-localized variations) are the primary reasons existing methods underperform and that the three proposed modules directly remedy this, but supplies no targeted evidence such as baseline failure-mode analysis, attention visualizations confirming SOC localization, or ablations isolating each module's contribution to those traits rather than generic gains; this causal link is load-bearing for the 'significant step' conclusion.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for highlighting issues with the abstract's claims. We will revise the abstract to include quantitative highlights and adjust the causal language to better align with the evidence presented in the full manuscript. Point-by-point responses follow.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that BatteryMFormer 'consistently outperforms state-of-the-art baselines' is presented without any quantitative results, specific baseline names, error metrics, error bars, or validation protocol, making the claim unverifiable from the provided evidence and load-bearing for the significance statement.

    Authors: We agree the abstract claim would be stronger with concrete numbers. In revision we will insert key quantitative results (e.g., average RMSE reduction of X% over baselines Y and Z across the four domains, with 5-run standard deviations) and a one-sentence validation protocol note, while staying within length limits. Full tables and protocols remain in Sections 4.1–4.2. revision: yes

  2. Referee: [Abstract] Abstract: The manuscript asserts that the two data characteristics (multi-level structure and SOC-localized variations) are the primary reasons existing methods underperform and that the three proposed modules directly remedy this, but supplies no targeted evidence such as baseline failure-mode analysis, attention visualizations confirming SOC localization, or ablations isolating each module's contribution to those traits rather than generic gains; this causal link is load-bearing for the 'significant step' conclusion.

    Authors: The manuscript already contains module ablations (Section 4.3) and attention visualizations aligned with SOC intervals (Figure 5). We did not include dedicated baseline failure-mode analysis. We will revise the abstract to state that the modules are designed to address the identified characteristics and that experiments demonstrate consistent gains, rather than asserting they are the 'primary reasons' for prior underperformance. This removes the stronger causal claim while preserving the motivation. revision: partial

Circularity Check

0 steps flagged

No significant circularity; claims rest on empirical validation

full rationale

The paper proposes an architecture motivated by two observed data traits and validates performance via experiments on four external battery datasets. No equations, derivations, or predictions appear that reduce by construction to fitted inputs or self-citations. The central claim (outperformance) is supported by baseline comparisons rather than any self-definitional loop, uniqueness theorem, or ansatz smuggled via prior work. This is the standard non-circular case for an applied ML architecture paper.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review performed on abstract only; no numerical parameters, mathematical axioms, or new postulated entities are described.

pith-pipeline@v0.9.1-grok · 5757 in / 1096 out tokens · 32253 ms · 2026-06-29T17:45:58.685525+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

62 extracted references · 37 canonical work pages · 5 internal anchors

  1. [1]

    2024.BatteryArchive.org.https://batteryarchive.org/index.html

    2024

  2. [2]

    Peter M Attia, Alexander Bills, Ferran Brosa Planella, Philipp Dechent, Goncalo Dos Reis, Matthieu Dubarry, Paul Gasper, Richard Gilchrist, Samuel Greenbank, David Howey, et al . 2022. “Knees” in lithium-ion battery aging trajectories. Journal of The Electrochemical Society169, 6 (2022), 060517

    2022

  3. [3]

    Peter M Attia, Aditya Grover, Norman Jin, Kristen A Severson, Todor M Markov, Yang-Hung Liao, Michael H Chen, Bryan Cheong, Nicholas Perkins, Zi Yang, et al. 2020. Closed-loop optimization of fast-charging protocols for batteries with machine learning.Nature578, 7795 (2020), 397–402

    2020

  4. [4]

    Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E. Hinton. 2016. Layer Normaliza- tion. arXiv:1607.06450 [stat.ML] https://arxiv.org/abs/1607.06450

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  5. [5]

    Birkl, Matthew R

    Christoph R. Birkl, Matthew R. Roberts, Euan McTurk, Peter G. Bruce, and David A. Howey. 2017. Degradation diagnostics for lithium ion cells.Journal of Power Sources341 (2017), 373–386. doi:10.1016/j.jpowsour.2016.12.011

    work page doi:10.1016/j.jpowsour.2016.12.011 2017

  6. [6]

    Mingyue Cheng, Jiqian Yang, Tingyue Pan, Qi Liu, Zhi Li, and Shijin Wang. 2025. ConvTimeNet: A Deep Hierarchical Fully Convolutional Model for Multivariate Time Series Analysis. InCompanion Proceedings of the ACM on Web Conference 2025(Sydney NSW, Australia)(WWW ’25). Association for Computing Machinery, New York, NY, USA, 171–180. doi:10.1145/3701716.3715214

    work page doi:10.1145/3701716.3715214 2025

  7. [7]

    Rose, Huada Lian, Alexis Geslin, Steven B

    Xiao Cui, Stephen Dongmin Kang, Sunny Wang, Justin A. Rose, Huada Lian, Alexis Geslin, Steven B. Torrisi, Martin Z. Bazant, Shijing Sun, and William C. KDD ’26, August 09–13, 2026, Jeju Island, Republic of Korea Ruifeng Tan et al. Chueh. 2024. Data-driven analysis of battery formation reveals the role of electrode utilization in extending cycle life.Joule...

    work page doi:10.1016/j.joule.2024.07.024 2026

  8. [8]

    Abhimanyu Das, Weihao Kong, Rajat Sen, and Yichen Zhou. 2024. A decoder-only foundation model for time-series forecasting. InProceedings of the 41st Interna- tional Conference on Machine Learning(Vienna, Austria)(ICML’24). JMLR.org, Article 404, 20 pages

    2024

  9. [9]

    Arnaud Devie, George Baure, and Matthieu Dubarry. 2018. Intrinsic Variability in the Degradation of a Batch of Commercial 18650 Lithium-Ion Cells.Energies 11, 5 (2018). doi:10.3390/en11051031

    work page doi:10.3390/en11051031 2018

  10. [10]

    Jakob Fleischmann, Mikael Hanicke, Evan Horetsky, Dina Ibrahim, Sören Jautelat, Martin Linder, Patrick Schaufuss, Lukas Torscht, and Alexandre van de Rijt. 2023. Battery 2030: Resilient, sustainable, and circular.McKinsey & Company16 (2023), 2023

    2023

  11. [11]

    Wei He, Nicholas Williard, Michael Osterman, and Michael Pecht. 2011. Prognos- tics of lithium-ion batteries based on Dempster–Shafer theory and the Bayesian Monte Carlo method.Journal of Power Sources196, 23 (2011), 10314–10321. doi:10.1016/j.jpowsour.2011.08.040

    work page doi:10.1016/j.jpowsour.2011.08.040 2011

  12. [12]

    Dan Hendrycks and Kevin Gimpel. 2016. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415(2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  13. [13]

    Boles, and Jean-Marie Tarascon

    Jiaqiang Huang, Steven T. Boles, and Jean-Marie Tarascon. 2022. Sensing as the key to battery lifetime and sustainability.Nature Sustainability5, 3 (2022), 194 –

    2022

  14. [14]

    http://dx.doi.org/10.1038/s41893-022-00859-y

    work page doi:10.1038/s41893-022-00859-y

  15. [15]

    Xinghao Huang, Chen Liang, Shengyu Tao, Yunhong Che, Ningyu Bian, Jiale Zhang, Runhua Wang, Yuqi Zhang, Bizhong Xia, and Xuan Zhang. 2026. IC2ML: Unified battery state-of-health, degradation trajectory and remaining useful life prediction via intra-cycle and inter-cycle enhanced machine learning.Journal of Power Sources666 (2026), 239148. doi:10.1016/j.jp...

    work page doi:10.1016/j.jpowsour.2025.239148 2026

  16. [16]

    Yaodi Huang, Pengcheng Zhang, Jiahuan Lu, Rui Xiong, and Zhongmin Cai. 2024. A transferable long-term lithium-ion battery aging trajectory prediction model considering internal resistance and capacity regeneration phenomenon.Applied Energy360 (2024), 122825. doi:10.1016/j.apenergy.2024.122825

    work page doi:10.1016/j.apenergy.2024.122825 2024

  17. [17]

    Jeevarajan, and Partha P

    Daniel Juarez-Robles, Saad Azam, Judith A. Jeevarajan, and Partha P. Mukherjee

  18. [18]

    Aging and Safety of Pouch Format Cells.Journal of The Electrochemical Society 168, 11 (nov 2021), 110501

    Degradation-Safety Analytics in Lithium-Ion Cells and Modules: Part III. Aging and Safety of Pouch Format Cells.Journal of The Electrochemical Society 168, 11 (nov 2021), 110501. doi:10.1149/1945-7111/ac30af

    work page doi:10.1149/1945-7111/ac30af 2021

  19. [19]

    Daniel Juarez-Robles, Judith A Jeevarajan, and Partha P Mukherjee. 2020. Degradation-safety analytics in lithium-ion cells: Part I. Aging under charge/discharge cycling.Journal of The Electrochemical Society167, 16 (2020), 160510

    2020

  20. [20]

    Minsoo Kim, Inwoo Kim, Jisub Kim, and Jang Wook Choi. 2023. Lifetime Predic- tion of Lithium Ion Batteries by Using the Heterogeneity of Graphite Anodes.ACS Energy Letters8, 7 (2023), 2946 – 2953. http://dx.doi.org/10.1021/acsenergylett. 3c00695

    work page doi:10.1021/acsenergylett 2023

  21. [21]

    Jinwen Li, Zhongwei Deng, Yunhong Che, Yi Xie, Xiaosong Hu, and Remus Teodorescu. 2024. Degradation Pattern Recognition and Features Extrapolation for Battery Capacity Trajectory Prediction.IEEE Transactions on Transportation Electrification10, 3 (2024), 7565–7579. doi:10.1109/TTE.2023.3336618

    work page doi:10.1109/tte.2023.3336618 2024

  22. [22]

    Juren Li, Yang Yang, Hanchen Su, Jiayu Liu, Youmin Chen, Jianfeng Zhang, and Lujia Pan. 2025. LiPM: Foundation Model for Lithium-Ion Battery Analysis. In Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2(Toronto ON, Canada)(KDD ’25). Association for Computing Machinery, New York, NY, USA, 1412–1423. doi:10.1145/37118...

    work page doi:10.1145/3711896.3737027 2025

  23. [23]

    Howey, and Chao Hu

    Tingkai Li, Zihao Zhou, Adam Thelen, David A. Howey, and Chao Hu. 2024. Predicting battery lifetime under varying usage conditions from early aging data. Cell Reports Physical Science5, 4 (2024), 101891. doi:10.1016/j.xcrp.2024.101891

    work page doi:10.1016/j.xcrp.2024.101891 2024

  24. [24]

    Weihan Li, Neil Sengupta, Philipp Dechent, David Howey, Anuradha Annaswamy, and Dirk Uwe Sauer. 2021. One-shot battery degradation trajectory prediction with deep learning.Journal of Power Sources506 (2021), 230024. doi:10.1016/j. jpowsour.2021.230024

    work page doi:10.1016/j 2021

  25. [25]

    Weihan Li, Haotian Zhang, Bruis van Vlijmen, Philipp Dechent, and Dirk Uwe Sauer. 2022. Forecasting battery capacity and power degradation with multi-task learning.Energy Storage Materials53 (2022), 453–466. doi:10.1016/j.ensm.2022. 09.013

    work page doi:10.1016/j.ensm.2022 2022

  26. [26]

    Peiyuan Liu, Beiliang Wu, Yifan Hu, Naiqi Li, Tao Dai, Jigang Bao, and Shu-Tao Xia. 2025. TimeBridge: Non-Stationarity Matters for Long-term Time Series Forecasting.International Conference on Machine Learning(2025)

    2025

  27. [27]

    Qingqiang Liu, Zhiqing Shang, Shixiang Lu, Yuanhong Liu, Yuchao Liu, and Sheng Yu. 2025. Physics-guided TL-LSTM network for early-stage degradation trajectory prediction of lithium-ion batteries.Journal of Energy Storage106 (2025), 114736. doi:10.1016/j.est.2024.114736

    work page doi:10.1016/j.est.2024.114736 2025

  28. [28]

    Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. 2024. iTransformer: Inverted Transformers Are Effective for Time Series Forecasting. InThe Twelfth International Conference on Learning Representations, ICLR 2024, Vienna, Austria, May 7-11, 2024. OpenReview.net, Vienna, Austria, 1–25. https://openreview.net/forum?id=J...

    2024

  29. [29]

    Guijun Ma, Songpei Xu, Benben Jiang, Cheng Cheng, Xin Yang, Yue Shen, Tao Yang, Yunhui Huang, Han Ding, and Ye Yuan. 2022. Real-time personalized health status prediction of lithium-ion batteries using deep transfer learning.Energy and Environmental Science15, 10 (2022), 4083 – 4094. http://dx.doi.org/10.1039/ d2ee01676a

    2022

  30. [30]

    Jinhao Meng, Lei Cai, Shengxiang Yang, Junxin Li, Feifan Zhou, Jichang Peng, and Zhengxiang Song. 2024. An Empirical-Informed Model for the Early Degra- dation Trajectory Prediction of Lithium-ion Battery.IEEE Transactions on Energy Conversion39, 4 (2024), 2299–2311. doi:10.1109/TEC.2024.3385093

    work page doi:10.1109/tec.2024.3385093 2024

  31. [31]

    Peyman Mohtat, Suhak Lee, Jason B Siegel, and Anna G Stefanopoulou. 2021. Reversible and irreversible expansion of lithium-ion batteries under a wide range of stress factors.Journal of The Electrochemical Society168, 10 (2021), 100520

    2021

  32. [32]

    Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam

    Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. 2023. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. InThe Eleventh International Conference on Learning Representations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023. OpenReview.net, Kigali, Rwanda, 1–24. https: //openreview.net/forum?id=Jbdc0vTOcol

    2023

  33. [33]

    Barkholtz, Armando Fresquez, Daniel L

    Yuliya Preger, Heather M. Barkholtz, Armando Fresquez, Daniel L. Campbell, Benjamin W. Juba, Jessica Romàn-Kustas, Summer R. Ferreira, and Babu Cha- lamala. 2020. Degradation of Commercial Lithium-Ion Cells as a Function of Chemistry and Cycling Conditions.Journal of The Electrochemical Society167, 12 (sep 2020), 120532. doi:10.1149/1945-7111/abae37

    work page doi:10.1149/1945-7111/abae37 2020

  34. [34]

    Lee, Dominik Linzner, Kathrin Michel, Leon Merker, Balazs B

    Fuzhan Rahmanian, Robert M. Lee, Dominik Linzner, Kathrin Michel, Leon Merker, Balazs B. Berkes, Leah Nuss, and Helge Soren Stein. 2024. Attention towards chemistry agnostic and explainable battery lifetime prediction.npj Computational Materials10, 1 (2024). http://dx.doi.org/10.1038/s41524-024- 01286-7

    work page doi:10.1038/s41524-024- 2024

  35. [35]

    Severson, Peter M

    Kristen A. Severson, Peter M. Attia, Norman Jin, Nicholas Perkins, Benben Jiang, Zi Yang, Michael H. Chen, Muratahan Aykol, Patrick K. Herring, Dimitrios Fraggedakis, Martin Z. Bazant, Stephen J. Harris, William C. Chueh, and Richard D. Braatz. 2019. Data-driven prediction of battery cycle life before capacity degra- dation.Nature Energy4, 5 (2019), 383 – 391

    2019

  36. [36]

    Quanyong Shen, Jian Li, Jiahao Nie, Zhengyi Bao, and Chenhan Wang. 2025. A Lightweight Multiscale Signal Learning Framework for Predicting Battery Degradation Trajectory.IEEE Sensors Journal25, 24 (2025), 44801 – 44812. http: //dx.doi.org/10.1109/JSEN.2025.3625630

    work page doi:10.1109/jsen.2025.3625630 2025

  37. [37]

    Ruifeng Tan, Weixiang Hong, Jia Li, Jiaqiang Huang, and Tong-Yi Zhang. 2025. Pretrained Battery Transformer (PBT): A battery life prediction foundation model. arXiv preprint arXiv:2512.16334(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  38. [38]

    Ruifeng Tan, Weixiang Hong, Jiayue Tang, Xibin Lu, Ruijun Ma, Xiang Zheng, Jia Li, Jiaqiang Huang, and Tong-Yi Zhang. 2025. BatteryLife: A Comprehensive Dataset and Benchmark for Battery Life Prediction. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2(Toronto ON, Canada)(KDD ’25). Association for Computing Machin...

    work page doi:10.1145/3711896.3737372 2025

  39. [39]

    Ruifeng Tan, Xibin Lu, Minhao Cheng, Jia Li, Jiaqiang Huang, and Tong-Yi Zhang

  40. [40]

    doi:10.1016/j.ensm

    Forecasting battery degradation trajectory under domain shift with domain generalization.Energy Storage Materials72 (2024), 103725. doi:10.1016/j.ensm. 2024.103725

    work page doi:10.1016/j.ensm 2024

  41. [41]

    Shuai Tan, Bin Ji, and Ye Pan. 2023. EMMN: Emotional Motion Memory Net- work for Audio-driven Emotional Talking Face Generation. In2023 IEEE/CVF International Conference on Computer Vision (ICCV). 22089–22099. doi:10.1109/ ICCV51070.2023.02024

    work page arXiv 2023

  42. [42]

    Peiwang Tang and Weitai Zhang. 2025. Unlocking the Power of Patch: Patch- Based MLP for Long-Term Time Series Forecasting.Proceedings of the AAAI Conference on Artificial Intelligence39, 12 (Apr. 2025), 12640–12648. doi:10.1609/ aaai.v39i12.33378

    2025

  43. [43]

    Shengyu Tao, Haizhou Liu, Chongbo Sun, Haocheng Ji, Guanjun Ji, Zhiyuan Han, Runhua Gao, Jun Ma, Ruifei Ma, Yuou Chen, et al . 2023. Collaborative and privacy-preserving retired battery sorting for profitable direct recycling via federated machine learning.Nature Communications14, 1 (2023), 8032

    2023

  44. [44]

    Shengyu Tao, Mengtian Zhang, Zixi Zhao, Haoyang Li, Ruifei Ma, Yunhong Che, Xin Sun, Lin Su, Chongbo Sun, Xiangyu Chen, Heng Chang, Shiji Zhou, Zepeng Li, Hanyang Lin, Yaojun Liu, Wenjun Yu, Zhongling Xu, Han Hao, Scott Moura, Xuan Zhang, Yang Li, Xiaosong Hu, and Guangmin Zhou. 2025. Non- destructive degradation pattern decoupling for early battery traje...

    work page doi:10.1039/d4ee03839h 2025

  45. [45]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need, In Advances in Neural Information Processing Systems.Advances in Neural Information Processing Systems2017-December, 5999 – 6009

    2017

  46. [46]

    Fujin Wang, Zhi Zhai, Zhibin Zhao, Yi Di, and Xuefeng Chen. 2024. Physics- informed neural network for lithium-ion battery degradation stable modeling and prognosis.Nature Communications15, 1 (2024), 4332. https://doi.org/10. 1038/s41467-024-48779-z

    2024

  47. [47]

    Shiyu Wang, Jiawei Li, Xiaoming Shi, Zhou Ye, Baichuan Mo, Wenze Lin, Sheng- tong Ju, Zhixuan Chu, and Ming Jin. 2025. TIMEMIXER++: A GENERAL TIME SERIES PATTERN MACHINE FOR UNIVERSAL PREDICTIVE ANALYSIS.13th International Conference on Learning Representations, ICLR 2025, 1662 – 1698. BatteryMFormer: Multi-level Learning for Battery Degradation Trajector...

    2025

  48. [48]

    Attia, Valentin Sulzer, Suhak Lee, Greg Less, and Anna Stefanopoulou

    Andrew Weng, Peyman Mohtat, Peter M. Attia, Valentin Sulzer, Suhak Lee, Greg Less, and Anna Stefanopoulou. 2021. Predicting the impact of formation protocols on battery lifetime immediately after manufacturing.Joule5, 11 (2021), 2971 –

    2021

  49. [49]

    http://dx.doi.org/10.1016/j.joule.2021.09.015

    work page doi:10.1016/j.joule.2021.09.015 2021

  50. [50]

    Jason Weston, Sumit Chopra, and Antoine Bordes. 2015. Memory Networks. arXiv:1410.3916 [cs.AI] https://arxiv.org/abs/1410.3916

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  51. [51]

    Ma, Kwok-Leung Tsui, and Michael Pecht

    Yinjiao Xing, Eden W.M. Ma, Kwok-Leung Tsui, and Michael Pecht. 2013. An ensemble model for predicting the remaining useful performance of lithium-ion batteries.Microelectronics Reliability53, 6 (2013), 811 – 820. http://dx.doi.org/10. 1016/j.microrel.2012.12.003

    2013

  52. [52]

    Jiexia Ye, Weiqi Zhang, Ziyue Li, Jia Li, and Fugee Tsung. 2025. MedSpaformer: a Transferable Transformer with Multi-granularity Token Sparsification for Medi- cal Time Series Classification. arXiv:2503.15578 [cs.LG] https://arxiv.org/abs/ 2503.15578

    work page arXiv 2025

  53. [53]

    Ailing Zeng, Muxi Chen, Lei Zhang, and Qiang Xu. 2023. Are Transformers Effective for Time Series Forecasting?. InThirty-Seventh AAAI Conference on Artificial Intelligence, AAAI 2023, Thirty-Fifth Conference on Innovative Applica- tions of Artificial Intelligence, IAAI 2023, Thirteenth Symposium on Educational Advances in Artificial Intelligence, EAAI 202...

    work page doi:10.1609/aaai.v37i9.26317 2023

  54. [54]

    Han Zhang, Xiaofan Gui, Shun Zheng, Ziheng Lu, Yuqi Li, and Jiang Bian. 2023. BATTERYML: AN OPEN-SOURCE PLATFORM FOR MACHINE LEARNING ON BATTERY DEGRADATION. (2023). http://dx.doi.org/10.48550/arXiv.2310.14714

    work page doi:10.48550/arxiv.2310.14714 2023

  55. [55]

    Han Zhang, Yuqi Li, Shun Zheng, Ziheng Lu, Xiaofan Gui, Wei Xu, and Jiang Bian. 2025. Battery lifetime prediction across diverse ageing conditions with inter-cell deep learning.Nature Machine Intelligence7, 2 (FEB 2025), 270–277. doi:10.1038/s42256-024-00972-x

    work page doi:10.1038/s42256-024-00972-x 2025

  56. [56]

    Jiawen Zhang, Shun Zheng, Wei Cao, Jiang Bian, and Jia Li. 2023. Warpformer: A Multi-scale Modeling Approach for Irregular Clinical Time Series. InProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (Long Beach, CA, USA)(KDD ’23). Association for Computing Machinery, New York, NY, USA, 3273–3285. doi:10.1145/3580305.3599543

    work page doi:10.1145/3580305.3599543 2023

  57. [57]

    Taolue Zhang, Ruifeng Tan, Pinxi Zhu, Tong-Yi Zhang, and Jiaqiang Huang. 2025. Unlocking Ultrafast Diagnosis of Retired Batteries via Interpretable Machine Learning and Optical Fiber Sensors.ACS Energy Letters10 (2025), 862–871

    2025

  58. [58]

    Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, and Jingren Zhou

  59. [59]

    Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models.arXiv preprint arXiv:2506.05176(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  60. [60]

    Haoyuan Zheng, Shaobin Yang, Weihua Xue, Shouzhen Xiao, Ding Shen, Wei Dong, and Xu Zhang. 2026. Self-discharge estimation for lithium-ion batteries based on formation data in production.Engineering Applications of Artificial Intelligence169 (2026), 114180. doi:10.1016/j.engappai.2026.114180

    work page doi:10.1016/j.engappai.2026.114180 2026

  61. [61]

    Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. 2021. Informer: Beyond Efficient Transformer for Long Se- quence Time-Series Forecasting.Proceedings of the AAAI Conference on Artificial Intelligence35, 12 (May 2021), 11106–11115. doi:10.1609/aaai.v35i12.17325

    work page doi:10.1609/aaai.v35i12.17325 2021

  62. [62]

    Jiangong Zhu, Yixiu Wang, Yuan Huang, R Bhushan Gopaluni, Yankai Cao, Michael Heere, Martin J Mühlbauer, Liuda Mereacre, Haifeng Dai, Xinhua Liu, et al. 2022. Data-driven capacity estimation of commercial lithium-ion batteries from voltage relaxation.Nature communications13, 1 (2022), 2261. A End-of-life Definition We define end-of-life𝑡eol as the first c...

    2022