Pith. sign in

REVIEW 4 major objections 5 minor 50 references

Multitask Battery Management with Flexible Pretraining

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims a single flexible pretrained masked autoencoder can replace task-specific battery-management models, outperforming them on all five tasks across eleven datasets.

desk verdict Useful pretraining framework for battery tasks, but the headline claims overstate what the data show. read the letter →

arxiv 2509.01323 v1 pith:3IPJTLW6 submitted 2025-09-01 cs.LG cs.AI

Hong Lu , Jiali Chen , Jingzhao Zhang , Guannan He , Xuebing Han , Minggao Ouyang This is my paper · ORCID
classification cs.LGcs.AI
keywords batterymanagementmaskedautoencoderpretrainingremainingusefullifestate-of-healthestimationanomalydetectionmissingdatatransferlearning
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper is trying to establish that a single pretraining-and-finetuning pipeline can replace the current practice of building a separate data-hungry model for each battery management task. It introduces the Flexible Masked Autoencoder (FMAE), which learns from battery data snippets that have different channel sets, and shows that after pretraining it outperforms task-specific baselines on cell-level capacity, internal resistance, system capacity, anomaly detection, and remaining-life prediction across eleven laboratory, electric-vehicle, and storage-system datasets. The flagship quantitative claim is that for remaining useful life, FMAE needs only two charge cycles per cell to match the error of a rival model that consumes 100 cycles. If true, the practical upshot is that one flexible model, tolerant of missing sensor channels, could serve as a common backbone for real-world battery management.

What carries the argument

The central object is the FMAE encoder-decoder with two additions: learnable channel tokens that replace masked or absent channels so the input format can vary across tasks and datasets, and embedded battery states (current, state of charge, mileage) inserted at masked decoder positions instead of vanilla position embeddings. Those state embeddings prevent the transformer from producing identical outputs for the same masked position across different snippets, and they encode time and usage context. The pretraining objective is masked reconstruction with both patch and channel masking; finetuning removes the decoder and attaches a linear head to averaged encoder features.

What would settle it

Take two snippets from the same or different cells with identical current, state-of-charge, and mileage values but different remaining signals, mask the same patches in both, and check whether the trained decoder reconstructs different outputs; if it produces nearly identical reconstructions, the conditioning is insufficient.

Watch

Extended reading notes

Core claim

FMAE adapts the masked autoencoder to multi-snippet, multi-channel battery data. During pretraining it randomly masks patches, whole channels, and cycles. Masked channels are padded with learnable channel tokens, which later stand in for missing channels at deployment. The decoder receives, at masked positions, embeddings of the snippet's current, state of charge, and mileage, which the paper argues prevents output collapse caused by identical mask tokens and supplies temporal context. After pretraining on six electric-vehicle datasets, the encoder is finetuned with a linear head per task. The paper reports consistent wins across all five tasks and eleven datasets, including a mean absolute

Load-bearing premise

The load-bearing premise is that a snippet's current, state of charge, and mileage carry enough information to tell apart different moments in battery life; if two different snippets show identical values for those three channels, the model has no way to tell them apart and the reconstruction task becomes ill-posed.

Editorial extensions

If this is right

  • Pretraining on unlabeled charge snippets can replace task-specific feature engineering: on capacity estimation FMAE beats random forest and XGBoost tuned with expert features, with tighter error spread across chemistries.
  • Remaining-life prediction becomes far cheaper in data: two cycles per cell instead of 100, so battery lifetime screening could happen early in a cell's life.
  • Missing channels are tolerable at deployment: removing system-level statistics still outperforms a full-channel LSTM, and single-voltage-channel capacity estimation is comparable to a hand-crafted voltage-relaxation feature method.
  • Pretraining itself contributes the gain: the same architecture trained only on downstream data loses roughly 3 to 16 percent across tasks.
  • One FMAE model can be finetuned to five task families across eleven datasets, suggesting a common backbone can aggregate battery data from lab, vehicle, and storage sources.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: the same mechanism, channel tokens plus state-conditioned decoding, could transfer to other dynamical systems with heterogeneous and partly missing sensor channels, such as fuel cells, electrolyzers, or power grids; the paper only gestures at this.
  • Editorial inference: if the two-cycle remaining-life result holds beyond the three lab datasets, fleet-level battery triage could be reordered, affecting warranty logistics and second-life battery grading.
  • Editorial inference: the conditioning design implies a testable boundary—if two snippets share identical current, state of charge, and mileage but differ in hidden degradation state, the decoder cannot distinguish them; conditioning on cycle index or an explicit capacity state would be a natural extension.
  • Editorial inference: pretraining uses only EV data, yet lab and storage finetuning still benefit; whether data diversity rather than data volume drives the gain is not isolated by the paper.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes FMAE, a masked-autoencoder-style pretraining framework for battery data that handles heterogeneous and missing channels via learnable channel tokens and captures correlations across time-separated snippets through embedded battery states. The authors pretrain on six EV datasets, then fine-tune for five battery management tasks (cell-level capacity and internal resistance estimation, system-level capacity estimation, anomaly detection, and RUL prediction) across eleven datasets, comparing against expert-feature baselines and deep-learning baselines under five-fold cross-validation. The headline claims are that FMAE consistently outperforms all task-specific methods and that RUL prediction uses 50 times less inference data while maintaining state-of-the-art accuracy.

Significance. If the claims held, the paper would make a useful contribution: a single pretrained representation that transfers across heterogeneous battery datasets and tasks, with released code/data, explicit handling of missing channels, and an ablation showing pretraining helps. The experimental design is generally sound—five-fold CV, multiple datasets, robustness checks, and an honest baseline-sweep in the supplement. However, the main claims as written are contradicted by the paper's own supplementary tables, so the significance as stated is not currently supported. The underlying method is promising; revision of the claims and reconciliation of the data-efficiency numbers could make the contribution publishable.

major comments (4)
  1. [Abstract and Results; Supplementary Table 2] The abstract's claim that 'FMAE consistently outperforms all task-specific methods across five battery management tasks with eleven battery datasets' is directly contradicted by Supplementary Table 2. For example: on MIT2 capacity estimation, RF (RMSE 0.27) and XGBoost (0.41) beat FMAE (0.47); on EV6 capacity, LSTM (1.90) beats FMAE (1.99); on EV6 anomaly detection, DyAD (94.05 AUROC) beats FMAE (85.19); on MIT1 RUL, Discharge (116 cycles RMSE) and BatLiNet (117) beat FMAE (129). The supplementary text itself acknowledges that 'some feature based methods achieved the best result on the specific task.' The central claim should be revised to a statement about average performance or to 'comparable or best on average,' and the paper should avoid overgeneralizing per-dataset superiority.
  2. [Results (RUL prediction), Methods, Supplementary Note 2, Supplementary Table 5] The RUL data-efficiency claim is internally inconsistent. The Abstract/Results state FMAE uses 2 cycles versus BatLiNet's 100 cycles, i.e., 50x less data. However, Methods state that FMAE 'takes two snippets separated by 20 cycles as input,' Supplementary Note 2 says both FMAE and LSTM use charging data from cycles 40, 60, 80, and 100 (four cycles), and Supplementary Table 5 compares FMAE (4 cycles) against BatLiNet (7 cycles). The '50 times less' ratio is not supported by any single consistent definition of cycle count in the paper. The authors must specify exactly which cycles/snippets are used by FMAE and by each baseline, and state the data-efficiency claim in terms of that protocol.
  3. [Methods (Decoder) and Supplementary Note 3] The decoder's use of embedded battery states (current, SoC, mileage) at masked positions is claimed to prevent output collapse and to make reconstruction well-posed. Supplementary Note 3 proves only that two identical mask tokens with identical vanilla position embeddings produce identical attention outputs. It does not show that the conditioning variables uniquely distinguish different snippets: two snippets from different cycles can share the same current, SoC, and mileage values at the masked positions, in which case the reconstruction target remains ambiguous. The paper should either provide a more precise argument for why these three channels are sufficient, or soften the claim that the design guarantees non-collapse beyond the specific duplicated-token case.
  4. [Results and Supplementary Table 2] The paper reports five-fold cross-validation but does not report variance or significance tests for the headline comparisons. The 'consistently outperforms all' claim is therefore stronger than the evidence supports, especially for datasets where FMAE loses to a baseline (e.g., MIT2 capacity, EV6 anomaly). Adding per-dataset error bars or a paired significance test across folds would allow the reader to judge whether the average improvements are meaningful, and would be needed to justify the superlative wording after the claims are rewritten.
minor comments (5)
  1. [Abstract/Results] The '50 times less inference data' phrase appears before the data-usage protocol is defined; consider moving the quantitative claim to a place where the cycle/snippet definitions have been introduced.
  2. [Results RUL and Supplementary Table 2] Supplementary Table 2 reports both FMAE (2 Cycles) and FMAE (1 Cycle), but Figure 4 and the main text only discuss the 2-cycle variant. Clarify whether the 1-cycle result is a sensitivity analysis or an alternative configuration.
  3. [Methods (Channel masking)] The sentence 'we sample a random subset S of [c] with cardinality cp channel masking' appears garbled—the probability p_channel_masking is not properly defined in the equation; please restate the sampling procedure cleanly.
  4. [Throughout] There are several typographical artifacts (e.g., 'V oltages', 'V ariance', '15 thousands works') that should be corrected in a final pass.
  5. [Supplementary Figure 2 caption] The caption's footnote '1 Since the EV3 dataset only contains single-digit abnormal EVs...' is a bit informal for a supplement; consider moving the dataset-exclusion rationale into the main text or a dedicated section.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FMAE's results are empirical and downstream predictions are not derived from the pretraining loss.

full rationale

The paper's core derivation is an empirical pretraining-finetuning pipeline, not a mathematical derivation from the model's own outputs. FMAE is pretrained with a reconstruction loss on unlabeled snippets; for downstream tasks the decoder is removed, a linear head is added, and the model is finetuned on held-out folds with task labels (Methods, 'FMAE finetuning and inference'). No downstream prediction is obtained by inverting the pretraining objective, and no fitted parameter is renamed as a prediction. The channel-masking and battery-state conditioning designs are motivated by the external MAE reference and by an explicit collapse argument in Supplementary Note 3, which shows that identical mask tokens produce identical attention outputs; this is a stated rationale rather than a self-citation or imported uniqueness theorem. Baselines such as Severson et al. (Nature Energy 2019) and BatLiNet (Zhang et al., Nat. Mach. Intell. 2025) are external comparisons. The only mild concern is that pretraining on the six EV datasets includes unlabeled snippets from vehicles later used in downstream EV evaluation, which is a potential evaluation-contamination issue, not a circular derivation: the pretraining loss does not encode the downstream labels and the fine-tuning predictions are not equivalent to the reconstruction targets. Internal inconsistencies such as the '50 times less' data-efficiency claim versus the four-cycle description in Methods are correctness/reporting issues, not circularity. Therefore the paper is not circular.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

No new physical entities, particles, forces, or conserved quantities are introduced. The model-level constructs (learnable channel tokens, battery state embeddings) are trainable parameters, not external postulates.

free parameters (4)
  • p_channel_masking = 0.4
    Probability of masking a channel during pretraining; chosen by hand (Supplementary Table 3). The design of learned channel tokens and the reported robustness depend on this value, but no sensitivity analysis is given.
  • p_patch_masking = 0.5
    Standard MAE masking ratio chosen by hand; affects reconstruction difficulty and downstream representation quality.
  • n snippets per group = 5
    Number of snippets per pretraining group; inter-snippet correlation modeling and the 'collapse' fix depend on this choice; no ablation.
  • fine-tuning learning rates = varies by task
    Per-task learning rates (e.g., 0.00625, 0.0625, 0.00625 for capacity on EV/ES/lab) were chosen by hand and are part of the method; results are reported for these specific values without sensitivity analysis.
assumptions (4)
  • domain assumption Masked reconstruction of randomly masked patches/channels produces transferable battery representations
    The entire pretraining framework assumes that self-supervised reconstruction of masked snippets yields features useful for downstream tasks; no theoretical justification, only empirical ablations (Fig. 5a).
  • domain assumption Battery data from laboratory and EV/BESS systems share common underlying electrochemical patterns that are transferable
    Pretraining on six EV datasets is assumed to generalize to lab datasets (MIT1, MIT2, KIT, THU) and BESS; this cross-domain transfer is the basis for RUL and lab capacity gains.
  • domain assumption Current, SoC, and mileage are sufficient battery state variables to disambiguate masked patches across snippets
    The decoder uses these three channels as conditioning to avoid collapsed predictions (Supplementary Note 3); the argument only proves collapse with identical inputs, not that these variables are sufficient for reconstruction.
  • standard math The softmax attention collapse argument is valid for multi-head attention
    Supplementary Note 3 derives identical outputs for single-head attention; the extension to multi-head is asserted but not proven, though it is plausible.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multitask Battery Management with Flexible Pretraining." pith.science (2026). https://pith.science/paper/3IPJTLW6

@misc{pith2026250901323,
  author       = {Pith},
  title        = {Pith review of: Multitask Battery Management with Flexible Pretraining},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3IPJTLW6}},
  note         = {Machine review of arXiv:2509.01323}
}
read the original abstract

Industrial-scale battery management involves various types of tasks, such as estimation, prediction, and system-level diagnostics. Each task employs distinct data across temporal scales, sensor resolutions, and data channels. Building task-specific methods requires a great deal of data and engineering effort, which limits the scalability of intelligent battery management. Here we present the Flexible Masked Autoencoder (FMAE), a flexible pretraining framework that can learn with missing battery data channels and capture inter-correlations across data snippets. FMAE learns unified battery representations from heterogeneous data and can be adopted by different tasks with minimal data and engineering efforts. Experimentally, FMAE consistently outperforms all task-specific methods across five battery management tasks with eleven battery datasets. On remaining life prediction tasks, FMAE uses 50 times less inference data while maintaining state-of-the-art results. Moreover, when real-world data lack certain information, such as system voltage, FMAE can still be applied with marginal performance impact, achieving comparable results with the best hand-crafted features. FMAE demonstrates a practical route to a flexible, data-efficient model that simplifies real-world multi-task management of dynamical systems.

Figures

Figures reproduced from arXiv: 2509.01323 by the authors.

Figure 1
Figure 1. Overview of the proposed FMAE method and its inference strategy. a. FMAE handles diverse data format input without structure modifications. Cell-level estimation utilizes three basic observations (voltage, current, and state of charge) to evaluate its current capacity and internal resistance. System-level estimation evaluates system capacity and detects anomalies by incorporating variations along cells. Predic￾tion … view at source ↗
Figure 2
Figure 2. The pretraining pipeline of the proposed method with missed channel modeling and inter￾snippet correlation capturing. a. To handle battery data heterogeneity, a portion of the patches and channels in the input data is masked and removed. The masked channels are padded with learnable tokens for maintaining data format during pretraining. b.To further capture inter-snippet correlation, the encoder is applied to the in… view at source ↗
Figure 3
Figure 3. Comparison of FMAE on cell-level and system-level estimation tasks. a. Cell-level capacity estimation and IR estimation are investigated using two evaluation metrics, absolute error and root mean square error (RMSE). The dotted points in the box plot indicate average SOH errors. The curly braces are the interquartile ranges (IQR) for each method. b. System-level capacity estimation and EV anomaly detection use absol… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Comparison of FMAE on RUL prediction task. RUL prediction covers two metrics: root-mean square error (RMSE) and mean absolute point error (MAPE). The grey dotted line indicates a naive baseline, using the average RUL of the training cells to predict the tested cells, d…
Figure 5
Figure 5. Figure 5: Improvements and robustness brought by our pretraining and finetuning strategies. a. Comparison of results with and without pretraining across four tasks. AUROC is employed as the evalua￾tion metric for anomaly detection, while RMSE is used for the other three tasks. T…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 47 canonical work pages

  1. [1]

    author Swift, M. W. , author Swift, J. W. & author Qi, Y. journal title Modeling the electrical double layer at solid-state electrochemical interfaces . Nature Computational Science volume 1 , pages 212--220 ( year 2021 )

    work page 2021

  2. [2]

    journal title Battery-powered freight trains

    author Zenith, F. journal title Battery-powered freight trains . Nature Energy volume 6 , pages 1003--1004 ( year 2021 )

    work page 2021

  3. [3]

    , author Zhao, X

    author Zhang, C. , author Zhao, X. , author Sacchi, R. & author You, F. journal title Trade-off between critical metal requirement and transportation decarbonization in automotive electrification . Nature Communications volume 14 , pages 1616 ( year 2023 )

    work page 2023

  4. [4]

    , author Hu, X

    author Zhang, H. , author Hu, X. , author Hu, Z. & author Moura, S. J. journal title Sustainable plug-in electric vehicle integration into power systems . Nature Reviews Electrical Engineering volume 1 , pages 35--52 ( year 2024 )

    work page 2024

  5. [5]

    , author Lill, F

    author Kittner, N. , author Lill, F. & author Kammen, D. M. journal title Energy storage deployment and innovation for the clean energy transition . Nature Energy volume 2 , pages 17125 ( year 2017 )

    work page 2017

  6. [6]

    , author Dan, Z

    author Song, A. , author Dan, Z. , author Zheng, S. & author Zhou, Y. journal title An electricity-driven mobility circular economy with lifecycle carbon footprints for climate-adaptive carbon neutrality transformation . Nature Communications volume 15 , pages 5905 ( year 2024 )

    work page 2024

  7. [7]

    , author Che, Y

    author Zhang, J. , author Che, Y. , author Teodorescu, R. , author Song, Z. & author Hu, X. journal title Energy storage management in electric vehicles . Nature Reviews Clean Technology volume 1 , pages 161--175 ( year 2025 )

    work page 2025

  8. [8]

    & author Willcox, K

    author Ferrari, A. & author Willcox, K. journal title Digital twins in mechanical and aerospace engineering . Nature Computational Science volume 4 , pages 178--183 ( year 2024 )

    work page 2024

Show all 50 references
  1. [9]

    author Thelen, A. et al. journal title Probabilistic machine learning for battery health diagnostics and prognostics—review and perspectives . npj Materials Sustainability volume 2 , pages 14 ( year 2024 )

    2024

  2. [10]

    , author Heinrich, F

    author von Bülow, F. , author Heinrich, F. & author Paxton, W. A. journal title The future of battery data and the state of health of lithium-ion batteries in automotive applications . Communications Engineering volume 3 , pages 173 ( year 2024 )

    2024

  3. [11]

    author Hu, J. et al. journal title Estimate state of charge in lithium-ion batteries with unknown data . Applied Energy volume 389 , pages 125736 ( year 2025 )

    2025

  4. [12]

    author Hong, J. et al. journal title Multi- forword-step state of charge prediction for real-world electric vehicles battery systems using a novel lstm-gru hybrid neural network . eTransportation volume 20 , pages 100322 ( year 2024 )

    2024

  5. [13]

    , author Ju, S

    author Li, P. , author Ju, S. , author Bai, S. , author Zhao, H. & author Zhang, H. journal title State of charge estimation for lithium-ion batteries based on physics-embedded neural network . Journal of Power Sources volume 640 , pages 236785 ( year 2025 )

    2025

  6. [14]

    author Li, Y. et al. journal title Random forest regression for online capacity estimation of lithium-ion batteries . Applied Energy volume 232 , pages 197--210 ( year 2018 )

    2018

  7. [15]

    author Tao, S. et al. journal title Generative learning assisted state-of-health estimation for sustainable battery recycling with random retirement conditions . Nature Communications volume 15 , pages 10154 ( year 2024 )

    2024

  8. [16]

    author Lu, Y. et al. journal title Towards real-world state of health estimation, part 1: Cell-level method using lithium-ion battery laboratory data . eTransportation volume 21 , pages 100338 ( year 2024 )

    2024

  9. [17]

    , author Strange, C

    author Ibraheem, R. , author Strange, C. & author Dos Reis, G. journal title Capacity and internal resistance of lithium-ion batteries: Full degradation curve prediction from voltage response at constant current at discharge . Journal of Power Sources volume 556 , pages 232477...

    2023

  10. [18]

    & author Dos Reis, G

    author Strange, C. & author Dos Reis, G. journal title Prediction of future capacity and internal resistance of li-ion cells from one cycle of input data . Energy and AI volume 5 , pages 100097 ( year 2021 )

    2021

  11. [19]

    , author San Martín, I

    author Pérez, A. , author San Martín, I. , author Sanchis, P. & author Ursúa, A. journal title A novel aging modeling approach for second-life lithium-ion batteries . eTransportation volume 24 , pages 100400 ( year 2025 )

    2025

  12. [20]

    author Severson, K. A. et al. journal title Data-driven prediction of battery cycle life before capacity degradation . Nature Energy volume 4 , pages 383--391 ( year 2019 )

    2019

  13. [21]

    author Zhang, H. et al. journal title Battery lifetime prediction across diverse ageing conditions with inter-cell deep learning . Nature Machine Intelligence volume 7 , pages 270--277 ( year 2025 )

    2025

  14. [22]

    author Zhang, J. et al. journal title Realistic fault detection of li-ion battery via dynamical deep learning . Nature Communications volume 14 , pages 5940 ( year 2023 )

    2023

  15. [23]

    author Lu, Y. et al. journal title A method of cell-to-cell variation evaluation for battery packs in electric vehicles with charging cloud data . eTransportation volume 6 , pages 100077 ( year 2020 )

    2020

  16. [24]

    , author Li, L

    author Xu, C. , author Li, L. , author Xu, Y. , author Han, X. & author Zheng, Y. journal title A vehicle-cloud collaborative method for multi-type fault diagnosis of lithium-ion batteries . eTransportation volume 12 , pages 100172 ( year 2022 )

    2022

  17. [25]

    author Cao, R. et al. journal title Model-constrained deep learning for online fault diagnosis in li-ion batteries over stochastic conditions . Nature Communications volume 16 , pages 1651 ( year 2025 )

    2025

  18. [26]

    , author Zhao, J

    author Ng, M.-F. , author Zhao, J. , author Yan, Q. , author Conduit, G. J. & author Seh, Z. W. journal title Predicting the state of charge and health of batteries using data-driven machine learning . Nature Machine Intelligence volume 2 , pages 161--170 ( year 2020 )

    2020

  19. [27]

    author Zhu, J. et al. journal title Data-driven capacity estimation of commercial lithium-ion batteries from voltage relaxation . Nature Communications volume 13 , pages 2261 ( year 2022 )

    2022

  20. [28]

    , author Xiong, R

    author Lu, J. , author Xiong, R. , author Tian, J. , author Wang, C. & author Sun, F. journal title Deep learning to estimate lithium-ion battery state of health without additional degradation experiments . Nature Communications volume 14 , pages 2760 ( year 2023 )

    2023

  21. [29]

    author Liu, H. et al. journal title Multi-modal framework for battery state of health evaluation using open-source electric vehicle data . Nature Communications volume 16 , pages 1137 ( year 2025 )

    2025

  22. [30]

    author Figgener, J. et al. journal title Multi-year field measurements of home storage systems and their use in capacity estimation . Nature Energy volume 9 , pages 1438--1447 ( year 2024 )

    2024

  23. [31]

    , author Boles, S

    author Huang, J. , author Boles, S. T. & author Tarascon, J.-M. journal title Sensing as the key to battery lifetime and sustainability . Nature Sustainability volume 5 , pages 194--204 ( year 2022 )

    2022

  24. [32]

    author Ma, G. et al. journal title Real-time personalized health status prediction of lithium-ion batteries using deep transfer learning . Energy & Environmental Science volume 15 , pages 4083--4094 ( year 2022 )

    2022

  25. [33]

    , author Williard, N

    author He, W. , author Williard, N. , author Osterman, M. & author Pecht, M. journal title Prognostics of lithium-ion batteries based on dempster–shafer theory and the bayesian monte carlo method . Journal of Power Sources volume 196 , pages 10314--10321 ( year 2011 )

    2011

  26. [34]

    , author Jeevarajan, J

    author Juarez-Robles, D. , author Jeevarajan, J. A. & author Mukherjee, P. P. journal title Degradation-safety analytics in lithium-ion cells: Part i. aging under charge/discharge cycling . Journal of The Electrochemical Society volume 167 , pages 160510 ( year 2020 )

    2020

  27. [35]

    author Li, W. et al. journal title One-shot battery degradation trajectory prediction with deep learning . Journal of Power Sources volume 506 , pages 230024 ( year 2021 )

    2021

  28. [36]

    author Raffel, C. et al. journal title Exploring the limits of transfer learning with a unified text-to-text transformer . Journal of Machine Learning Research volume 21 , pages 1--67 ( year 2020 )

    2020

  29. [37]

    author Dosovitskiy, A. et al. title An image is worth 16x16 words: Transformers for image recognition at scale . In booktitle International Conference on Learning Representations ( year 2021 )

    2021

  30. [38]

    author Xu, H. et al. journal title A whole-slide foundation model for digital pathology from real-world data . Nature volume 630 , pages 181--188 ( year 2024 )

    2024

  31. [39]

    author Merchant, A. et al. journal title Scaling deep learning for materials discovery . Nature volume 624 , pages 80--85 ( year 2023 )

    2023

  32. [40]

    author Trinh, T. H. , author Wu, Y. , author Le, Q. V. , author He, H. & author Luong, T. journal title Solving olympiad geometry without human demonstrations . Nature volume 625 , pages 476--482 ( year 2024 )

    2024

  33. [41]

    author Theodoris, C. V. et al. journal title Transfer learning enables predictions in network biology . Nature volume 618 , pages 616--624 ( year 2023 )

    2023

  34. [42]

    journal title GB/T 32960.3-2016 Technical specifications of remote service and management system for electric vehicles - Part 3: Communication protocol and data format

    author General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China & author National Standardization Administration . journal title GB/T 32960.3-2016 Technical specifications of remote service and management system for electric ve...

    2016

  35. [43]

    author International Electrotechnical Commission . journal title IEC 61851-24:2023 Electric vehicle conductive charging system - Part 24: Digital communication between a DC EV supply equipment and an electric vehicle for control of DC charging . IEC Standard ( year 2023 )

    2023

  36. [44]

    author He, K. et al. journal title Masked autoencoders are scalable vision learners . In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition pages 16000--16009 ( year IEEE, 2022 )

    2022

  37. [45]

    author Attia, P. M. et al. journal title Closed-loop optimization of fast-charging protocols for batteries with machine learning . Nature volume 578 , pages 397--402 ( year 2020 )

    2020

  38. [46]

    author Vaswani, A. et al. journal title Attention is all you need . Advances in Neural Information Processing Systems volume 30 , pages 5998--6008 ( year 2017 )

    2017

  39. [47]

    , " * write output.state after.block = add.period write newline

    ENTRY address archive author booktitle chapter edition editor eprint howpublished institution journal key month note number organization pages publisher school series title type url doi volume year label INTEGERS output.state before.all mid.sentence after.sentence after.block ...

  40. [48]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

  41. [49]

    author Pedregosa, F. et al. journal title Scikit-learn: Machine learning in python . Journal of Machine Learning Research volume 12 , pages 2825--2830 ( year 2011 )

    2011

  42. [50]

    & author Guestrin, C

    author Chen, T. & author Guestrin, C. title XGBoost : A scalable tree boosting system . In booktitle Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , KDD '16, pages 785--794 , 10.1145/2939672.2939785 ( publisher ACM , address...

    2016

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.