pith. sign in

arxiv: 2606.11605 · v1 · pith:P2MY4YODnew · submitted 2026-06-10 · 💻 cs.LG · cs.AI

Physics-Distilled Neural Network enabled by Large Language Models for Manufacturing Process-Property Predictive Modeling

Ge Song , Kiarash Naghavi Khanghah , Anandkumar Patel , Rajiv Malhotra , Hongyi Xu This is my paper

Pith reviewed 2026-06-27 10:25 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords knowledge distillationlarge language modelsmanufacturingphysics priorspredictive modelingneural networksgraph attentiondata scarcity
0
0 comments X

The pith

Distilling physics knowledge extracted by large language models into lightweight neural networks enables accurate predictions of manufacturing process outcomes despite limited experimental data.

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

The paper develops a framework that uses large language models to pull analytical physics principles from existing literature and incorporate them into a teacher neural network. This teacher then transfers the knowledge to a simpler student model that can make fast predictions. The approach is tested on five manufacturing processes and shows it can handle cases where the extracted physics information is incomplete. A special graph-masked attention layer in the teacher helps model the relationships between process inputs. The result is a system that achieves high accuracy in data-poor situations and supports real-time industrial monitoring.

Core claim

The central discovery is a physics-distilled neural network framework where LLM-derived analytical priors serve as privileged knowledge in a teacher model equipped with a Graph-Masked Attention layer, which is then distilled into a lightweight student predictor that maintains high accuracy and achieves inference frequencies exceeding 6000 Hz across diverse manufacturing processes, demonstrating robustness to suboptimal priors.

What carries the argument

The knowledge distillation pipeline that integrates LLM-extracted physics priors into a privileged teacher model using Graph-Masked Attention, transferred to a student predictor.

If this is right

  • Accurate predictions become possible with fewer experiments in manufacturing.
  • The student model supports real-time edge deployment on standard hardware.
  • Performance holds across multiple process types even with imperfect physics priors.
  • The method provides a way to combine literature knowledge with machine learning for industrial use.

Where Pith is reading between the lines

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

  • Similar techniques might help in other fields where physical literature exists but data is scarce.
  • The fault tolerance reduces the risk of relying on LLM extractions.
  • It could lead to hybrid models that improve over time with more data.

Load-bearing premise

Large language models can reliably extract accurate analytical physics priors from scientific literature to guide the teacher model.

What would settle it

Observing a substantial drop in predictive accuracy for the student model when the LLM-derived priors are removed or when applied to a new manufacturing process not included in the five tested domains.

Figures

Figures reproduced from arXiv: 2606.11605 by Anandkumar Patel, Ge Song, Hongyi Xu, Kiarash Naghavi Khanghah, Rajiv Malhotra.

Figure 1
Figure 1. Figure 1: FIGURE 1 [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIGURE 2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIGURE 3 [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
read the original abstract

Predicting process-property relationships in manufacturing is often challenged by high experimental costs and the limited interpretability of complex 'black-box' models. This paper proposes a novel knowledge distillation framework designed to achieve high-accuracy predictions in data-scarce scenarios. The framework integrates analytical physics priors, which are systematically extracted from scientific literature via Large Language Models, into a privileged teacher model. We employ a Graph-Masked Attention layer to capture the complex physical dependencies among input variables showing strict setpoints or a combination of static and high-frequency temporal signatures. This privileged knowledge is distilled into a lightweight student predictor for inference. The feasibility and robustness of the framework are evaluated through a comprehensive experiment across five diverse manufacturing processes. To ensure statistical reliability, given the small dataset sizes, a repeated K-fold cross-validation technique is employed to quantify model stability and generalization. Results indicate that the proposed framework consistently achieves high predictive accuracy across all evaluated domains. Most importantly, the architecture demonstrates significant fault tolerance by maintaining robust predictive performance even in scenarios where LLM-derived analytical priors are suboptimal or incomplete. Furthermore, the student predictor achieves an inference frequency exceeding 6000 Hz, which facilitates real-time edge deployment on standard industrial hardware. This work provides a scalable solution for bridging the gap between theoretical physics and real-time industrial monitoring in data-limited environments.

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

0 major / 3 minor

Summary. The paper proposes a knowledge distillation framework for manufacturing process-property prediction in data-scarce settings. LLM-extracted analytical physics priors are incorporated into a privileged teacher model equipped with a Graph-Masked Attention layer to model physical dependencies among inputs (setpoints or mixed static/temporal signatures); this knowledge is then distilled to a lightweight student model. Feasibility is assessed across five manufacturing processes via repeated K-fold cross-validation, with reported outcomes of consistently high predictive accuracy, explicit fault tolerance to suboptimal or incomplete LLM priors, and student inference exceeding 6000 Hz for real-time edge deployment.

Significance. If the empirical results hold, the work offers a practical route to embed physics knowledge into deployable predictors without requiring large datasets, while the demonstrated fault tolerance to imperfect priors and the high inference rate address key barriers in industrial monitoring. The approach scales prior extraction via LLMs and separates teacher (physics-informed) from student (lightweight) stages, which could generalize to other domains with expensive experimentation.

minor comments (3)
  1. [Abstract] The abstract states that repeated K-fold cross-validation was used 'to ensure statistical reliability' but provides no values for the number of repeats, folds, or resulting variance estimates; adding these numbers (and a table of per-process metrics) would strengthen the generalization claim.
  2. The Graph-Masked Attention mechanism is described only at a high level; a short methods subsection or pseudocode showing how the mask is constructed from the physics priors would clarify whether the layer enforces the claimed 'strict setpoints or ... temporal signatures' dependencies.
  3. No baseline comparisons (pure data-driven models, physics-only models, or alternative distillation methods) are mentioned in the provided summary; including at least one standard baseline per process would make the accuracy and fault-tolerance gains easier to interpret.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary, recognition of the framework's practical value in data-scarce settings, and recommendation for minor revision. The assessment of fault tolerance, inference speed, and potential generalization is appreciated.

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents an empirical framework that extracts physics priors via LLMs, incorporates them into a Graph-Masked Attention teacher, and distills to a student model, with performance quantified via repeated K-fold cross-validation on five manufacturing processes. No load-bearing derivation, equation, or claim reduces by construction to its own inputs or to a self-citation chain; the reported fault tolerance to suboptimal priors is framed as an observed experimental outcome rather than a logical necessity derived from the priors themselves. The central results therefore remain falsifiable against external data and do not exhibit any of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the unverified ability of LLMs to extract usable analytical physics priors and on the assumption that the distillation step preserves both accuracy and fault tolerance; no free parameters or invented entities are explicitly listed in the abstract.

axioms (1)
  • domain assumption Large language models can systematically extract accurate analytical physics priors from scientific literature
    This is required for the teacher model to receive privileged knowledge.

pith-pipeline@v0.9.1-grok · 5779 in / 1266 out tokens · 28388 ms · 2026-06-27T10:25:35.915223+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

65 extracted references · 1 canonical work pages

  1. [1]

    Laser-based selective material processing for next-generation additive manufacturing,

    H. Park, J. J. Park, P.-D. Bui, H. Yoon, C. P. Grigoropoulos, D. Lee and S. H. Ko, "Laser-based selective material processing for next-generation additive manufacturing," Advanced Materials, vol. 36, no. 34, p. 2307586, 2024

    2024

  2. [2]

    An additive manufacturing approach for fabrication of agarose hydrogel structures for protein sorption application,

    F. Cao, H. N. Zadeh, K. Swiacka, J. Maculewicz, D. Bowles, T. Huber and D. Clucas, "An additive manufacturing approach for fabrication of agarose hydrogel structures for protein sorption application," Materials & Design, vol. 250, p. 113581, 2025

    2025

  3. [3]

    The Role of Calibration Engineering In Strengthening Reliability Of US Advanced Manufacturing Systems Through Artificial Intelligence,

    E. A. Haque, "The Role of Calibration Engineering In Strengthening Reliability Of US Advanced Manufacturing Systems Through Artificial Intelligence," Review of Applied Science and Technology, vol. 4, no. 02, pp. 820 - 851, 2025

    2025

  4. [4]

    Innovative Data Models: Transforming Material Process Design and Optimization,

    A. M. Horr, M. Hartmann and F. Haunreiter, "Innovative Data Models: Transforming Material Process Design and Optimization," Metals, vol. 15, no. 8, p. 873, 2025

    2025

  5. [5]

    Modelling the influence of UV curing strategies for optimisation of inkjet based 3D printing,

    P. Zhao, Y . He, G. F. Trindade, M. Baumers, D. J. Irvine, R. J. Hague, I. A. Ashcroft and R. D. Wildman, "Modelling the influence of UV curing strategies for optimisation of inkjet based 3D printing," Materials & Design, vol. 208, p. 109889, 2021

    2021

  6. [6]

    The economics of additive manufacturing: Towards a general cost model including process failure,

    J. Ding, M. Baumers, E. A. Clark and R. D. Wildman, "The economics of additive manufacturing: Towards a general cost model including process failure," International Journal of Production Economics, vol. 237, p. 108087, 2021

    2021

  7. [7]

    Sustainability aspects of additive manufacturing: Leveraging resource efficiency via product design optimization and laser powder bed fusion,

    P. Nyamekye, R. Lakshmanan, V . Tepponen and S. Westman, "Sustainability aspects of additive manufacturing: Leveraging resource efficiency via product design optimization and laser powder bed fusion," Heliyon, vol. 10, no. 1, 2024

    2024

  8. [8]

    A review of conventional versus additive manufacturing for metals: life-cycle environmental and economic analysis,

    A. Mecheter, F. Tarlochan and M. Kucukvar, "A review of conventional versus additive manufacturing for metals: life-cycle environmental and economic analysis," Sustainability, vol. 15, no. 16, p. 12299, 2023

    2023

  9. [9]

    Large language models for extrapolative modeling of manufacturing processes,

    K. Naghavi Khanghah, A. Patel, R. Malhotra and H. Xu, "Large language models for extrapolative modeling of manufacturing processes," Journal of Intelligent Manufacturing, pp. 1-29, 2025

    2025

  10. [10]

    Enhanced Magnet -aided Laser Induced Plasma Micromachining (E-MLIP) for Expanded Geometric Capabilities,

    R. Malhotra, A. Patel, K. N. Khanghah and H. Xu, "Enhanced Magnet -aided Laser Induced Plasma Micromachining (E-MLIP) for Expanded Geometric Capabilities," CIRP Annals, vol. 74, no. 1, pp. 257 -261, 2025

    2025

  11. [11]

    A Print -Form-Sinter process for manufacturing 3D electronics on complex surfaces,

    R. Malhotra, M. N. Jahangir and H. Xu, "A Print -Form-Sinter process for manufacturing 3D electronics on complex surfaces," CIRP Annals, 2026

    2026

  12. [12]

    Physics- informed machine learning in intelligent manufacturing: a review,

    J. Leng, K. Zuo, C. Xu, X. Zhou, S. Zheng, J. Kang, Q. Liu, X. Chen, W. Shen, L. Wang and others, "Physics- informed machine learning in intelligent manufacturing: a review," Journal of Intelligent Manufacturing, pp. 1- 43, 2025

    2025

  13. [13]

    Improving hydrothermal carbonization prediction by machine learning: towards a more accurate and less expensive process modelling through data augmentation,

    B. Cosenza, A. Picone, M. V olpe and A. Messineo, "Improving hydrothermal carbonization prediction by machine learning: towards a more accurate and less expensive process modelling through data augmentation," Energy, p. 137373, 2025

    2025

  14. [14]

    Predicting plant stress using SAM-L: novel self-adaptive-meta learner with XAI based on soil moisture and chlorophyll analysis,

    T. Alsanoosy and J. A. Malik, "Predicting plant stress using SAM-L: novel self-adaptive-meta learner with XAI based on soil moisture and chlorophyll analysis," Scientific Reports, vol. 15, no. 1, p. 42304, 2025

    2025

  15. [15]

    A massive MIMO channel Estimation method based on hybrid deep learning model with regularization techniques,

    X. Tian and Q. Zheng, "A massive MIMO channel Estimation method based on hybrid deep learning model with regularization techniques," International Journal of Intelligent Systems, vol. 2025, no. 1, p. 2597866, 2025

    2025

  16. [16]

    Reconstruction and generation of porous metamaterial units via variational graph autoencoder and large language model,

    K. Naghavi Khanghah, Z. Wang and H. Xu, "Reconstruction and generation of porous metamaterial units via variational graph autoencoder and large language model," Journal of Computing and Information Science in Engineering, vol. 25, no. 2, p. 021003, 2025

    2025

  17. [17]

    MCERF: Advancing Multimodal LLM Evaluation of Engineering Documentation with Enhanced Retrieval,

    K. N. Khanghah, H. A. Nguyen, A. C. Doris, A. M. Vahedi, D. Grandi, F. Ahmed and H. Xu, "MCERF: Advancing Multimodal LLM Evaluation of Engineering Documentation with Enhanced Retrieval," arXiv preprint arXiv:2604.09552, 2026

    Pith/arXiv arXiv 2026

  18. [18]

    Retrieval -augmented generation for knowledge -intensive nlp tasks,

    P. Lewis, E. Perez, A. Piktus, F. Petroni, V . Karpukhin, N. Goyal, H. Kuttler, M. Lewis, W.-t. Yih, T. Rocktaschel and others, "Retrieval -augmented generation for knowledge -intensive nlp tasks," Advances in neural information processing systems, vol. 33, pp. 9459-9474, 2020

    2020

  19. [19]

    RAG in health care: a novel framework for improving communication and decision-making by addressing LLM limitations,

    K. K. Y . Ng, I. Matsuba and P. C. Zhang, "RAG in health care: a novel framework for improving communication and decision-making by addressing LLM limitations," Nejm Ai, vol. 2, no. 1, p. AIra2400380, 2025

    2025

  20. [20]

    Large language models for information retrieval: A survey,

    Y . Zhu, H. Yuan, S. Wang, J. Liu, W. Liu, C. Deng, H. Chen, Z. Liu, Z. Dou and J. -R. Wen, "Large language models for information retrieval: A survey," ACM Transactions on Information Systems, vol. 44, no. 1, pp. 1 - 54, 2025

    2025

  21. [21]

    Human--artificial intelligence teaming for scientific information extraction from data -driven additive manufacturing literature using large language models,

    M. Safdar, J. Xie, A. Mircea and Y . F. Zhao, "Human--artificial intelligence teaming for scientific information extraction from data -driven additive manufacturing literature using large language models," Journal of Computing and Information Science in Engineering, vol. 25, no. 7, p. 074501, 2025

    2025

  22. [22]

    Designqa: A multimodal benchmark for evaluating large language models’ understanding of engineering documentation,

    A. C. Doris, D. Grandi, R. Tomich, M. F. Alam, M. Ataei, H. Cheong and F. Ahmed, "Designqa: A multimodal benchmark for evaluating large language models’ understanding of engineering documentation," Journal of Computing and Information Science in Engineering, vol. 25, no. 2, p. 021009, 2025

    2025

  23. [23]

    Knowledge extraction for additive manufacturing process via named entity recognition with LLMs,

    X. Liu, J. A. Erkoyuncu, J. Y . H. Fuh, W. F. Lu and B. Li, "Knowledge extraction for additive manufacturing process via named entity recognition with LLMs," Robotics and Computer-Integrated Manufacturing, vol. 93, p. 102900, 2025

    2025

  24. [24]

    AMGPT: A large language model for contextual querying in additive manufacturing,

    A. Chandrasekhar, J. Chan, F. Ogoke, O. Ajenifujah and A. B. Farimani, "AMGPT: A large language model for contextual querying in additive manufacturing," Additive Manufacturing Letters, vol. 11, p. 100232, 2024

    2024

  25. [25]

    Ontology-based retrieval augmented generation (rag) for genai- supported additive manufacturing,

    Y . Park, P. Witherell, N. A. Surovi and H. Cho, "Ontology-based retrieval augmented generation (rag) for genai- supported additive manufacturing," 2024

    2024

  26. [26]

    Knowledge Graph Fusion with Large Language Models for Accurate, Explainable Manufacturing Process Planning,

    D. Hoang, D. Gorsich, M. P. Castanier and F. Imani, "Knowledge Graph Fusion with Large Language Models for Accurate, Explainable Manufacturing Process Planning," arXiv preprint arXiv:2506.13026, 2025

    arXiv 2025

  27. [27]

    AdditiveLLM: Large language models predict defects in metals additive manufacturing,

    P. Pak and A. B. Farimani, "AdditiveLLM: Large language models predict defects in metals additive manufacturing," Additive Manufacturing Letters, vol. 14, p. 100292, 2025

    2025

  28. [28]

    Integrating large language models (LLMs) into printing process optimization for binder jetting printed Ti-6Al-4V ,

    Y . Qian, X. Luo, Q. Wei, S. Yi, L. Zhang, D. Huang, N. Qin, R. Liu, R. Huang and K. O. a. o. Yu, "Integrating large language models (LLMs) into printing process optimization for binder jetting printed Ti-6Al-4V ," Powder Technology, vol. 474, p. 122295, 2026

    2026

  29. [29]

    Llm -sr: Scientific equation discovery via programming with large language models,

    P. Shojaee, K. Meidani, S. Gupta, A. B. Farimani and C. K. Reddy, "Llm -sr: Scientific equation discovery via programming with large language models," arXiv preprint arXiv:2404.18400, 2024

    arXiv 2024

  30. [30]

    Large language models for automatic equation discovery of nonlinear dynamics,

    M. Du, Y . Chen, Z. Wang, L. Nie and D. Zhang, "Large language models for automatic equation discovery of nonlinear dynamics," Physics of Fluids, vol. 36, no. 9, 2024

    2024

  31. [31]

    Physics-guided, physics-informed, and physics-encoded neural networks and operators in scientific computing: Fluid and solid mechanics,

    S. A. Faroughi, N. M. Pawar, C. Fernandes, M. Raissi, S. Das, N. K. Kalantari and S. Kourosh Mahjour, "Physics-guided, physics-informed, and physics-encoded neural networks and operators in scientific computing: Fluid and solid mechanics," Journal of Computing and Information Science in Engineering, vol. 24, no. 4, p. 040802, 2024

    2024

  32. [32]

    Physics -informed machine learning across manufacturing processes: Recent advances, challenges, and directions,

    D. Ra, J. Lee, M. Lee, S. Kwak, S. Lee and S. Lee, "Physics -informed machine learning across manufacturing processes: Recent advances, challenges, and directions," Journal of Manufacturing Systems, vol. 85, pp. 72-95, 2026

    2026

  33. [33]

    Physics -guided neural operator for data-driven composites manufacturing process modelling,

    G. Chen, Y . Li, X. Liu, C. Mehdi-Souzani, Q. Meng, J. Zhou and X. Hao, "Physics -guided neural operator for data-driven composites manufacturing process modelling," Journal of Manufacturing Systems, vol. 70, pp. 217- 229, 2023

    2023

  34. [34]

    Physics guided neural network for machining tool wear prediction,

    J. Wang, Y . Li, R. Zhao and R. X. Gao, "Physics guided neural network for machining tool wear prediction," Journal of Manufacturing Systems, vol. 57, pp. 298-310, 2020

    2020

  35. [35]

    Physics-guided neural network-based feedforward control for seamless pipe manufacturing process,

    L. Filipovic, L. Milicic, M. Ristanovic, V . Dimitrijevic and P. Jovanovic, "Physics-guided neural network-based feedforward control for seamless pipe manufacturing process," Applied Sciences, vol. 15, no. 4, p. 2229, 2025

    2025

  36. [36]

    Embedding physical knowledge in deep neural networks for predicting the phonon dispersion curves of cellular metamaterials,

    Z. Wang, W. Xian, Y . Li and H. Xu, "Embedding physical knowledge in deep neural networks for predicting the phonon dispersion curves of cellular metamaterials," Computational Mechanics, vol. 72, no. 1, pp. 221 -239, 2023

    2023

  37. [37]

    Enhancing weld line visibility prediction in injection molding using physics -informed neural networks,

    A. Pieressa, G. Baruffa, M. Sorgato and G. Lucchetta, "Enhancing weld line visibility prediction in injection molding using physics -informed neural networks," Journal of Intelligent Manufacturing, vol. 36, no. 6, pp. 4305-4318, 2025

    2025

  38. [38]

    Prediction of 3D temperature field through single 2D temperature data based on transfer learning -based PINN model in laser -based directed energy deposition,

    S. Peng, S. Yang, B. Gao, W. Liu, F. Wang and Z. Tang, "Prediction of 3D temperature field through single 2D temperature data based on transfer learning -based PINN model in laser -based directed energy deposition," Journal of Manufacturing Processes, vol. 138, pp. 140-156, 2025

    2025

  39. [39]

    Physics -informed neural networks in materials modeling and design: a review,

    A. Michaloglou, I. Papadimitriou, I. Gialampoukidis, S. Vrochidis and I. Kompatsiaris, "Physics -informed neural networks in materials modeling and design: a review," Archives of Computational Methods in Engineering, pp. 1-38, 2025

    2025

  40. [40]

    Physics -Informed Machine Learning in Design and Manufacturing: Status and Challenges,

    L. Pan, G. Li, T. Zhu, D. Liu, Y . Wang and Y . Lu, "Physics -Informed Machine Learning in Design and Manufacturing: Status and Challenges," Journal of Computing and Information Science in Engineering, vol. 25, no. 12, p. 120804, 2025

    2025

  41. [41]

    Distilling the knowledge in a neural network,

    G. Hinton, O. Vinyals and J. Dean, "Distilling the knowledge in a neural network," arXiv preprint arXiv:1503.02531, 2015

    Pith/arXiv arXiv 2015

  42. [42]

    A systematic review of knowledge distillation in industrial predictive maintenance: Applications, methods and challenges,

    L. Wang, J. K. Chaw, M. C. Ang, X. Cheng, H. B. Zaman, S. S. Gunasekaran and M. A. Mahmoud, "A systematic review of knowledge distillation in industrial predictive maintenance: Applications, methods and challenges," ICT Express, 2025

    2025

  43. [43]

    Advanced fault diagnosis in milling cutting tools using vision transformers with semi -supervised learning and uncertainty quantification,

    M. F. Siddique, M. Umar, W. Ahmad and J.-M. Kim, "Advanced fault diagnosis in milling cutting tools using vision transformers with semi -supervised learning and uncertainty quantification," Scientific reports, vol. 15, no. 1, p. 42460, 2025

    2025

  44. [44]

    Knowledge distillation-based information sharing for online process monitoring in decentralized manufacturing system,

    Z. Shi, Y . Li and C. Liu, "Knowledge distillation-based information sharing for online process monitoring in decentralized manufacturing system," Journal of Intelligent Manufacturing, vol. 36, no. 3, pp. 2177-2192, 2025

    2025

  45. [45]

    Automatic network structure discovery of physics informed neural networks via knowledge distillation,

    Z. Liu, Y . Liu, X. Yan, W. Liu, H. Nie, S. Guo and C. -a. Zhang, "Automatic network structure discovery of physics informed neural networks via knowledge distillation," Nature Communications, vol. 16, no. 1, p. 9558, 2025

    2025

  46. [46]

    Real-Time Model Predictive Control of Monoclonal Antibody Capture in Continuous Manufacturing Using Physics -Informed Neural Networks Accelerated Mechanistic Modeling,

    S.-Y . Tang, Y .-H. Yuan, Y .-N. Sun, W.-H. Pan, S.-J. Yao and D.-Q. Lin, "Real-Time Model Predictive Control of Monoclonal Antibody Capture in Continuous Manufacturing Using Physics -Informed Neural Networks Accelerated Mechanistic Modeling," Biotechnology and Bioengineering, 2025

    2025

  47. [47]

    A new learning paradigm: Learning using privileged information,

    V . Vapnik and A. Vashist, "A new learning paradigm: Learning using privileged information," Neural networks, vol. 22, no. 5-6, pp. 544-557, 2009

    2009

  48. [48]

    text -embedding-3-small,

    OpenAI, "text -embedding-3-small," OpenAI, [Online]. Available: https://platform.openai.com/docs/guides/embeddings.. [Accessed 15 May 2026]

    2026

  49. [49]

    GPT-4o-mini,

    OpenAI, "GPT-4o-mini," OpenAI, [Online]. Available: https://openai.com/index/gpt -4o-mini-advancing-cost- efficient-intelligence/. [Accessed 15 May 2026]

    2026

  50. [50]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser and I. Polosukhin, "Attention is all you need," Advances in neural information processing systems, vol. 30, 2017

    2017

  51. [51]

    Injection Molding Dataset (Viscometer, Euromap77, Quality Table),

    T. Grunemann and J. Schüder, "Injection Molding Dataset (Viscometer, Euromap77, Quality Table)," 1 January

  52. [52]

    Available: https://doi.org/10.23728/b2share.d8502ea56c544e069ebda44c3edd441b

    [Online]. Available: https://doi.org/10.23728/b2share.d8502ea56c544e069ebda44c3edd441b.. [Accessed 19 May 2026]

    work page doi:10.23728/b2share.d8502ea56c544e069ebda44c3edd441b 2026

  53. [53]

    Turning of 42CrMo4+ QT under different scenarios: Dataset of machining, roughness and residual stress,

    D. Diaz-Salamanca, S. Alvarez, M. Muniz -Calvente, P. Ebrahimzadeh, I. Llavori, A. Zabala, P. Pando, C. S. Alvarez, I. Fernandez-Pariente and M. a. o. Larranaga, "Turning of 42CrMo4+ QT under different scenarios: Dataset of machining, roughness and residual stress," Data in Brief, vol. 56, p. 110793, 2024

    2024

  54. [54]

    Experimental investigation and optimization of laser induced plasma micromachining using flowing water,

    X. Wang, Y . Huang, X. Wang, B. Xu, J. Feng and B. Shen, "Experimental investigation and optimization of laser induced plasma micromachining using flowing water," Optics & Laser Technology, vol. 126, p. 106067, 2020

    2020

  55. [55]

    The effects of process parameters on tensile characteristics and printing time for masked stereolithography components, analyzed using the response surface method,

    A. Temiz, "The effects of process parameters on tensile characteristics and printing time for masked stereolithography components, analyzed using the response surface method," Journal of Materials Engineering and Performance, vol. 33, no. 18, pp. 9356-9365, 2024

    2024

  56. [56]

    Multi -response optimization of the turn -assisted deep cold rolling process parameters for enhanced surface characteristics and residual stress of AISI 4140 steel shafts,

    P. Prabhu, S. Kulkarni and S. Sharma, "Multi -response optimization of the turn -assisted deep cold rolling process parameters for enhanced surface characteristics and residual stress of AISI 4140 steel shafts," Journal of Materials Research and Technology, vol. 9, no. 5, pp. 11402-11423, 2020

    2020

  57. [57]

    Surface properties and corrosion behavior of turn -assisted deep-cold-rolled AISI 4140 steel,

    P. Prabhu, D. Prabhu, S. Sharma and S. Kulkarni, "Surface properties and corrosion behavior of turn -assisted deep-cold-rolled AISI 4140 steel," Journal of materials engineering and performance, vol. 29, no. 9, pp. 5871- 5885, 2020

    2020

  58. [58]

    Learning to extrapolate: A transductive approach,

    A. Netanyahu, A. Gupta, M. Simchowitz, K. Zhang and P. Agrawal, "Learning to extrapolate: A transductive approach," arXiv preprint arXiv:2304.14329, 2023

    arXiv 2023

  59. [59]

    Random forests,

    L. Breiman, "Random forests," Machine learning, vol. 45, no. 1, pp. 5-32, 2001

    2001

  60. [60]

    Xgboost: A scalable tree boosting system,

    T. Chen and C. Guestrin, "Xgboost: A scalable tree boosting system," in Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, 2016

    2016

  61. [61]

    Multilayer perceptron (MLP),

    H. Taud and J. -F. Mas, "Multilayer perceptron (MLP)," in Geomatic approaches for modeling land change scenarios, 2017

    2017

  62. [62]

    Knowledge distillation: A survey,

    J. Gou, B. Yu, S. J. Maybank and D. Tao, "Knowledge distillation: A survey," International journal of computer vision, vol. 129, no. 6, pp. 1789-1819, 2021

    2021

  63. [63]

    The application of physics-informed machine learning in multiphysics modeling in chemical engineering,

    Z. Wu, H. Wang, C. He, B. Zhang, T. Xu and Q. Chen, "The application of physics-informed machine learning in multiphysics modeling in chemical engineering," Industrial & Engineering Chemistry Research, vol. 62, no. 44, pp. 18178-18204, 2023. Appendix

    2023

  64. [64]

    Because some datasets are strictly limited in size, the hidden layer dimensions are intentionally kept lightweight to prevent severe overfitting

    Detailed Neural Network Architecture To ensure reproducibility , the detailed network architectures and layer dimensions for both the Privileged Teacher and the Student Predictor across all five manufacturing processes are provided below. Because some datasets are strictly limited in size, the hidden layer dimensions are intentionally kept lightweight to ...

  65. [65]

    For complete form, please refer to [9]

    Knowledge Retrieval Query and Model Generation Prompt In this section a summarized Query Form used in RAG model to extract parametric knowledge from the literature and a summarized Prompt Form used to translate that knowledge into analytical Python models with iterative refinement has been included. For complete form, please refer to [9]. The query is sub...