Pith. sign in

REVIEW 4 major objections 6 minor 36 references

Wafer Defect Root Cause Analysis with Partial Trajectory Regression

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A new method scores each wafer process step by the change it causes in predicted defect density, and the scores sum exactly to the final prediction.

desk verdict Useful representation-learning idea for wafer trajectories, but the causal attribution claim and the one-wafer validation don't support the paper's central assertion. read the letter →

arxiv 2507.20357 v1 pith:DVNG3TVI submitted 2025-07-27 cs.LG

Kohei Miyaguchi , Masao Joko , Rebekah Sheraw , Tsuyoshi Idé This is my paper · ORCID
classification cs.LG
keywords waferdefectrootcauseanalysispartialtrajectoryregressionprocessembeddingattributionscoresemiconductormanufacturingcounterfactualreasoningadditive
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

This paper tries to make wafer defect root-cause analysis systematic by turning each process step along a wafer's history into an interpretable score. The framework, Partial Trajectory Regression (PTR), predicts defect density from variable-length process trajectories, then defines the attribution of step $k$ as the difference in prediction between the trajectory up to $k$ and the trajectory up to $k-1$. The authors claim this difference quantifies the causal intervention of that step, and that the scores obey an additive property, so cumulative scores trace how defects accumulate along a route. They demonstrate on real fab history data that process embeddings learned from string similarity capture tool and recipe relations, and that the cumulative attribution plot highlights long waiting times as likely defect sources. If this holds, engineers would get a principled, step-by-step account of why a wafer came out defective, without relying solely on manual ad hoc analysis.

What carries the argument

The load-bearing object is the constrained recurrent cell $z_k = \psi(t_k,t_{k-1})x_k + z_{k-1}$ (Eq. 6), which reduces a variable-length process trajectory to a weighted sum of process embeddings and thereby lets any prediction function $f$ accept partial trajectories as input. The attribution identity $\alpha_k(\xi)=f(z_k)-f(z_{k-1})$ (Eq. 9) compares two counterfactual inputs—the trajectory with and without step $k$—and the additive property (Theorem VI.1) guarantees the step scores sum to the full change in prediction. Two named representation modules support this: proc2vec, a kernel embedding of process tokens constructed from a substring kernel over equipment, recipe, tool type, and photo layer attributes, and route2vec, the mapping from the sequence of these vectors to $z_k$. The recurrence's linear form is what makes the additive attribution exact; if the cell were nonlinear or included interaction terms, Eq. (9) would no longer decompose cleanly.

What would settle it

Compare PTR's attribution scores against a controlled perturbation on a held-out wafer: set step $k$'s embedding to zero (or perturb its timestamp) and measure whether the actual change in predicted defect density equals $\alpha_k$; a systematic mismatch would falsify the claim that Eq. (9) quantifies the step's causal intervention.

Watch

Extended reading notes

Core claim

The central claim is that the attribution score $\alpha_k(\xi)=f(z_k)-f(z_{k-1})$ (the paper's Eq. 9) quantifies the causal intervention of process step $k$ under a potential-outcome interpretation, where $f$ is the learned prediction function and $z_k$ is the recurrent embedding of the partial trajectory up to step $k$. Since the trajectory representation is built with the recurrence $z_k=\psi(t_k,t_{k-1})x_k+z_{k-1}$ (Eq. 6), the difference between consecutive predictions is the model's marginal response to adding step $k$ in its actual temporal position. The paper proves that these scores telescope: $\sum_{l=1}^k \alpha_l = f(z_k)-f(z_0)$, making cumulative attribution plots meaningful for single-wafer diagnosis. On a real front-end-of-line dataset of 787 wafers spanning hundreds of processes, the kernel-based embedding correlates with defect density substantially better than constant or one-hot encodings, and the cumulative attribution plot identifies long waiting times at tools as candidate root causes.

Load-bearing premise

The load-bearing premise is the additive, context-free recurrence $z_k=\psi(t_k,t_{k-1})x_k+z_{k-1}$, which assumes each process step's contribution is independent of its neighbors, so that if steps truly interact, the attribution score conflates interaction effects with the step's own causal contribution.

Editorial extensions

If this is right

  • Each process step in a wafer's history receives a numerical score that adds up to the model's final defect-density prediction, enabling cumulative attribution plots that show how defects accumulate along a route.
  • The framework handles variable-length routes and heterogeneous processes without fixed-dimensional feature engineering, leveraging embedding similarity among similar tools and recipes to cope with small effective sample sizes.
  • Because the recurrence is additive in the process embeddings, the attribution score equals the marginal prediction change from inserting step $k$ at its temporal position, which the paper interprets as that step's causal intervention.
  • On the real FEOL dataset, the method produced attribution jumps at unusually long waiting times, suggesting that waiting durations are candidate root causes of high defect density and actionable targets for process optimization.
  • The prediction and attribution modules are separable, so the same additive attribution identity can be applied to any prediction function trained on the partial-trajectory representation.

Reading between the lines

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

  • The same additive-trajectory attribution scheme could be applied to other sequential manufacturing or logistics processes, wherever a step's contribution is represented as a sum over a route.
  • Because the recurrence omits interaction terms, the method will attribute interaction effects to whichever step appears later in the trajectory; a synthetic experiment with known interacting steps could quantify how much misattribution that causes.
  • The long-wait-time finding suggests that the temporal mapping $\psi$ conveys signal beyond mere presence of a step; extending it beyond $\log_{10}(1+\cdot)$ to a learned function of wait duration may sharpen attribution.
  • A stronger causal reading would require adjusting for unobserved confounders, which the current potential-outcome interpretation silently assumes; the framework as presented measures the model's response to step inclusion, not necessarily the physical intervention effect.
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 / 6 minor

Summary. The paper proposes Partial Trajectory Regression (PTR), a framework for wafer defect root cause analysis on variable-length process trajectories. Processes are embedded with a string-kernel method called proc2vec; trajectories are encoded by a constrained RNN with cell z_k = psi(t_k, t_{k-1}) x_k + z_{k-1}; and a linear MLP maps the encoded state to defect density. Attribution for process k is defined as alpha_k(ξ) = f(z_k) - f(z_{k-1}) and is claimed to quantify the causal intervention of process k under Rubin's potential-outcome framework. The paper reports prediction correlations of 0.27, 0.52, 0.61, and 0.87 for different embedding/model configurations on 787 wafers from the NY CREATES fab, and presents a cumulative attribution plot for a single wafer with two highlighted jumps, A and B, said to correspond to unusually long waiting times.

Significance. If the attribution scores were validated against known root causes, PTR would be a practically valuable interpretable tool for semiconductor root cause analysis. The paper has useful ingredients: a string-kernel embedding that shares information across similar tools and recipes, a partial-trajectory representation that handles variable-length routes, and a real fab dataset. However, the central claim of causal attribution is not established: the score is a pathwise additive decomposition, not a counterfactual effect, and the empirical evaluation contains no ground truth for attribution, no baseline attribution method, and only one uncontrolled case study. The significance is therefore potential rather than demonstrated.

major comments (4)
  1. [Section VI, Eq. (9)] The attribution score alpha_k(ξ) = f(z_k) - f(z_{k-1}) is presented as a 'causal intervention' following Rubin's potential-outcome framework, but it is not a counterfactual comparison of two trajectories. Under Eq. (6), z_k = psi(t_k, t_{k-1}) x_k + z_{k-1}, so the difference is the marginal effect of appending step k to the prefix, not the effect of removing step k from the full route; the genuinely counterfactual trajectory without k would have different later states and timestamps. Theorem VI.1 is a tautological telescoping sum and does not provide causal identification. The causal claim requires assumptions (e.g., no unobserved confounding, no interference between steps) that are neither stated nor defended. The causal interpretation should be removed or supported by a formal identification argument and empirical validation.
  2. [Section V, Eq. (6)] The cell z_k = psi(t_k, t_{k-1}) x_k + z_{k-1} assumes that process effects are additive and independent of the surrounding context. This is a load-bearing modeling assumption: if processes interact, for example a lithography step only causes defects when a prior etch leaves a specific surface, then the difference f(z_k) - f(z_{k-1}) conflates interaction effects and does not isolate the contribution of step k. No evidence or diagnostic is provided for this assumption, and no comparison with a more expressive model (even one with pairwise interaction terms) is made. At minimum, the assumption should be stated explicitly and tested, for example on a synthetic dataset with known interacting effects.
  3. [Section VII] The empirical evaluation does not validate the attribution scores. Figure 4 shows a single held-out wafer, and the highlighted jumps A and B are interpreted as long waiting times without engineering verification, a statistical test, or comparison with other wafers or baseline attribution methods such as Shapley values. The reported correlation coefficients (0.27, 0.52, 0.61, 0.87) measure predictive fit, not attribution accuracy. In addition, Eq. (8) trains the model to predict the final defect density y^{(n)} from every partial representation z_k, which is not a well-defined prediction target for partial trajectories; this needs justification. The central claim of 'demonstrated effectiveness' currently rests on an uncontrolled visual example.
  4. [Sections V and VII] The manuscript omits essential experimental details: 'details on training are left to a longer version of the paper in preparation' (Section V), 'Further details are omitted here for brevity' (Section VII), and reference [4] lists page 'TBD'. The exact process-token attributes, hyperparameters, train/test split, number of distinct processes, and the procedure for selecting the displayed wafer are not provided. Without these, the reported results cannot be reproduced or independently checked, which is especially problematic because the main evidence for the attribution claim is a single case study.
minor comments (6)
  1. [Section VI, proof of Theorem VI.1] The displayed proof contains a typo: the first term reads f(x_k) - f(z_{k-1}) but should be f(z_k) - f(z_{k-1}).
  2. [Eq. (8)] The inner summation uses the same final defect density y^{(n)} as the target for every partial prefix z_k; please clarify whether y is measured after the full trajectory and why partial-prefix predictions should be trained against it.
  3. [Figure 3 caption] The caption states that the color mapping and axis scales are undisclosed, which prevents the reader from interpreting the clustering; please provide a legend and explain the scaling choices.
  4. [Section IV] The sentence beginning 'hence, for cross-process attribution' starts with a lowercase letter after a period; please fix this typographical error.
  5. [Reference [4]] The page field is listed as 'TBD'; if the work is forthcoming, please provide the final page numbers or a DOI.
  6. [Figure 2 caption] The caption says the projection layer is the only learnable component, but Eq. (7) is an MLP and Eq. (8) includes an L1 penalty on its parameters; please clarify which components (embedding, recurrent cell, projection) are learned and which are frozen.

Circularity Check

2 steps flagged · score 6.0 of 10

Attribution score is a within-trajectory additive decomposition; the causal interpretation and the effectiveness demonstration both reduce to the definition of the score.

  1. self definitional [Section VI, Eq. (9) and Theorem VI.1]
    "To quantify the influence of the k-th process, we ask: Is there a significant difference in the prediction outcome between the partial trajectories z_k and z_{k−1}? This comparison between two counterfactual inputs measures the potential outcome when including process k. ... Following Rubin's potential outcome framework [35], the attribution module quantifies the causal intervention of process k as: αk(ξ) = f(z_k) − f(z_{k−1}). (9)"

    The causal attribution is not derived from a counterfactual or potential-outcome model; it is defined as the difference between two consecutive internal states of the same observed trajectory. Given the recurrent cell z_k = ψ(t_k,t_{k−1}) x_k + z_{k−1} (Eq. 6), α_k(ξ) is by construction f(z_{k−1} + ψ x_k) − f(z_{k−1}), a pathwise additive increment of the fitted prediction function, not the effect of a hypothetical trajectory that omits process k. The true counterfactual without k would contain the remaining processes with adjusted timestamps, and its encoded state would not be z_{k−1}. The 'Additive Property' (Eq. 10) is a telescoping identity of this definition, not an independent result. Thus the causal claim reduces to the definition of the score.

  2. fitted input called prediction [Section VII, evaluation of Fig. 4]
    "Finally, Fig. 4 presents the cumulative attribution score, which plots [sum_i α_i + f(z_0)] (11) for a specific wafer from the held-out dataset at each timestamp τ... Notable jumps, labeled A and B, are highlighted in the plot. Upon further inspection, these jumps correspond to unusually long waiting times at certain tools, suggesting potential root causes of high defect density. This example demonstrates how PTR effectively identifies problematic processes, providing actionable insights for RCA."

    The 'demonstration of effectiveness' is a visual reading of the model's own cumulative attribution score. By Eq. (10), the plotted quantity equals f(z_k), the fitted prediction function's output for the prefix, so the highlighted jumps are features of the fitted model itself. No independent ground truth, no external confirmation of the suspected root causes, no baseline attribution method, and no ablation is used to validate the attribution. The model output is therefore presented as its own validation: the effectiveness claim reduces to the output of the fitted function, i.e., the fitted prediction is renamed as a defect-cause score.

full rationale

The forward model is not circular: the embedding (proc2vec) and the regression module (route2vec plus f) are trained against measured defect density y, and the predictive correlation is an external quantity. The circularity lies in the attribution step. Equation (9) defines the causal attribution score as a difference of the fitted function at two consecutive states of the same trajectory; the Rubin potential-outcome interpretation is asserted, not derived, and the additive property is a tautology of that definition. The empirical section then validates the attribution by pointing at jumps in the model's own cumulative score without any independent outcome, making the effectiveness claim self-referential. There are self-citations (e.g., trajectory regression [14], [15]), but they are not load-bearing for the central attribution claim; the central reduction is by definition and by the self-referential evaluation.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The framework rests on several domain assumptions about which process attributes are informative and on the additive recurrence that yields the neat attribution property. The causal interpretation is an additional assumption, not a derived consequence.

free parameters (5)
  • Embedding dimensionality D = Not specified.
    Hyperparameter in Eq. (4) that controls the proc2vec representation; no value or tuning procedure is reported.
  • Regularization strength nu = Not specified.
    Hyperparameter in Eq. (8) for the L1 penalty; no value is reported.
  • Time-mapping function psi = log10(1+·)
    Chosen by hand without justification; it weights how waiting times affect process contributions in Eq. (6).
  • Substring kernel parameters = Not specified.
    The custom variant of the substring kernel [33,34] used to compute the kernel matrix K is not described in detail, so its hyperparameters are unknown.
  • Additional process features = Unspecified.
    The correlation improvement to 0.87 in Section VII relies on 'additional process features and optimizing the prediction model', but these features and optimization choices are not disclosed.
assumptions (4)
  • domain assumption The process token formed by concatenating eqp, recipe, tool_type, photo_layer, route (Eq. 3) captures all relevant process information for defect prediction.
    This assumes high-level attributes suffice and tool traces are irrelevant, which is not empirically justified.
  • ad hoc to paper The trajectory representation evolves as a cumulative sum of weighted process embeddings (Eq. 6), i.e., process effects are additive and independent of context.
    This recurrence is chosen to enable the additive attribution property; no empirical support for additivity is given.
  • domain assumption The learned function f trained on observed trajectories can be used to evaluate counterfactual partial trajectories.
    Attribution Eq. (9) treats f(z_k) and f(z_{k-1}) as potential outcomes without stating causal identifiability conditions such as sequential ignorability.
  • domain assumption Wafers can be treated as independent samples despite lot- and batch-based processing.
    Section III acknowledges non-independence, but the loss in Eq. (8) treats the N pairs as independent.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Wafer Defect Root Cause Analysis with Partial Trajectory Regression." pith.science (2026). https://pith.science/paper/DVNG3TVI

@misc{pith2026250720357,
  author       = {Pith},
  title        = {Pith review of: Wafer Defect Root Cause Analysis with Partial Trajectory Regression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DVNG3TVI}},
  note         = {Machine review of arXiv:2507.20357}
}
read the original abstract

Identifying upstream processes responsible for wafer defects is challenging due to the combinatorial nature of process flows and the inherent variability in processing routes, which arises from factors such as rework operations and random process waiting times. This paper presents a novel framework for wafer defect root cause analysis, called Partial Trajectory Regression (PTR). The proposed framework is carefully designed to address the limitations of conventional vector-based regression models, particularly in handling variable-length processing routes that span a large number of heterogeneous physical processes. To compute the attribution score of each process given a detected high defect density on a specific wafer, we propose a new algorithm that compares two counterfactual outcomes derived from partial process trajectories. This is enabled by new representation learning methods, proc2vec and route2vec. We demonstrate the effectiveness of the proposed framework using real wafer history data from the NY CREATES fab in Albany.

Figures

Figures reproduced from arXiv: 2507.20357 by the authors.

Figure 1
Figure 1. Causal analysis framework with partial trajectory regression (PTR). [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. System architecture of the PTR framework. Only the projection layer [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Distribution of the proc2vec embeddings for three wafer samples, visualized with t-SNE (the perplexity is set to be 30). The color coding corresponds to distinction among the equipment types including Wet Process, Rapid Thermal Processing, Inspection, Lithography, Reactive Ion Etching, Ion Implantation, Furnace, Chemical Mechanical Polishing, although the exact mapping as well as axis scales are undisclosed. N = 787… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 31 canonical work pages

  1. [4]

    Sequence-aware inline measurement attribution for good-bad wafer diagnosis,

    K. Miyaguchi, M. Joko, R. Sheraw, and T. Id ´e, “Sequence-aware inline measurement attribution for good-bad wafer diagnosis,” in2025 SEMI Advanced Semiconductor Manufacturing Conference (ASMC), p. TBD, IEEE, 2025

    work page 2025

  2. [1]

    Multi-step virtual metrology for semiconductor manufacturing: A multilevel and regularization methods-based approach,

    G. A. Susto, S. Pampuri, A. Schirru, A. Beghi, and G. De Nico- lao, “Multi-step virtual metrology for semiconductor manufacturing: A multilevel and regularization methods-based approach,”Computers & Operations Research, vol. 53, pp. 328–337, 2015

    work page 2015

  3. [2]

    Virtual metrology on semiconductor manufacturing based on just-in- time learning,

    M. A. Jebri, E. El Adel, G. Graton, M. Ouladsine, and J. Pinaton, “Virtual metrology on semiconductor manufacturing based on just-in- time learning,”IFAC-PapersOnLine, vol. 49, no. 12, pp. 89–94, 2016

    work page 2016

  4. [3]

    Variable selection under missing values and unlabeled data in semiconductor pro- cesses,

    K.-J. Kim, K.-J. Kim, C.-H. Jun, I.-G. Chong, and G.-Y . Song, “Variable selection under missing values and unlabeled data in semiconductor pro- cesses,”IEEE Transactions on Semiconductor Manufacturing, vol. 32, no. 1, pp. 121–128, 2018

    work page 2018

  5. [5]

    Soft-sensing conformer: A curriculum learning-based convolutional transformer,

    J. Yella, C. Zhang, S. Petrov, Y . Huang, X. Qian, A. A. Minai, and S. Bom, “Soft-sensing conformer: A curriculum learning-based convolutional transformer,” in2021 IEEE International Conference on Big Data (Big Data), pp. 1990–1998, IEEE, 2021

    work page 1990

  6. [6]

    Deep learning-based virtual metrology in multivariate time series,

    S. Han, et al., “Deep learning-based virtual metrology in multivariate time series,” in2023 IEEE International Conference on Prognostics and Health Management (ICPHM), pp. 30–37, IEEE, 2023

    work page 2023

  7. [7]

    Deep learning-based sequence modeling for advanced process control in semiconductor manufacturing,

    F. Dalla Zuanna, N. Gentner, and G. A. Susto, “Deep learning-based sequence modeling for advanced process control in semiconductor manufacturing,”IFAC-PapersOnLine, vol. 56, no. 2, pp. 8744–8751, 2023

    work page 2023

  8. [8]

    Recurrent feature-incorporated convolutional neural network for virtual metrology of the chemical mechanical pla- narization process,

    K. B. Lee and C. O. Kim, “Recurrent feature-incorporated convolutional neural network for virtual metrology of the chemical mechanical pla- narization process,”Journal of Intelligent Manufacturing, vol. 31, no. 1, pp. 73–86, 2020

    work page 2020

Show all 36 references
  1. [9]

    Virtual metrology of material removal rate using a one-dimensional convolutional neural network-based bidirec- tional long short-term memory network with attention,

    C.-Y . Hsu and Y .-W. Lu, “Virtual metrology of material removal rate using a one-dimensional convolutional neural network-based bidirec- tional long short-term memory network with attention,”Computers & Industrial Engineering, vol. 186, p. 109701, 2023

    2023

  2. [10]

    Machine learning assisted new product setup,

    J. A. Torres, I. Kissiov, M. Essam, C. Hartig, R. Gardner, K. Jantzen, S. Schueler, and M. Niehoff, “Machine learning assisted new product setup,” in2020 31st Annual SEMI Advanced Semiconductor Manufac- turing Conference (ASMC), pp. 1–5, IEEE, 2020

    2020

  3. [11]

    Using explainable artificial intelligence to improve process quality: evidence from semiconductor manufacturing,

    J. Senoner, T. Netland, and S. Feuerriegel, “Using explainable artificial intelligence to improve process quality: evidence from semiconductor manufacturing,”Management Science, vol. 68, no. 8, pp. 5704–5723, 2022

    2022

  4. [12]

    An expandable yield prediction framework using explainable artificial intelligence for semiconductor manufacturing,

    Y . Lee and Y . Roh, “An expandable yield prediction framework using explainable artificial intelligence for semiconductor manufacturing,” Applied Sciences, vol. 13, no. 4, p. 2660, 2023

    2023

  5. [13]

    Enhanced yield prediction in semiconductor manufacturing: Innovative strategies for imbalanced sample management and root cause analysis,

    P. Guo and Y . Chen, “Enhanced yield prediction in semiconductor manufacturing: Innovative strategies for imbalanced sample management and root cause analysis,” in2024 IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits (IPFA), pp. 1–6, IEEE, 2024

    2024

  6. [14]

    Travel-time prediction using gaussian process regression: A trajectory-based approach,

    T. Id ´e and S. Kato, “Travel-time prediction using gaussian process regression: A trajectory-based approach,” inProceedings of the 2009 SIAM International Conference on Data Mining, pp. 1185–1196, SIAM, 2009

    2009

  7. [15]

    Trajectory regression on road networks,

    T. Id ´e and M. Sugiyama, “Trajectory regression on road networks,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 25, pp. 203–208, 2011

    2011

  8. [16]

    Root cause analysis in industrial manufacturing: A scoping review of current research, challenges and the promises of ai-driven approaches,

    D. Pietsch, M. Matthes, U. Wieland, S. Ihlenfeldt, and T. Munkelt, “Root cause analysis in industrial manufacturing: A scoping review of current research, challenges and the promises of ai-driven approaches,”Journal of Manufacturing and Materials Processing, vol. 8, no. 6, p. ...

    2024

  9. [17]

    A fast ramp-up framework for wafer yield improvement in semiconductor manufacturing systems,

    H.-W. Xu, Q.-H. Zhang, Y .-N. Sun, Q.-L. Chen, W. Qin, Y .-L. Lv, and J. Zhang, “A fast ramp-up framework for wafer yield improvement in semiconductor manufacturing systems,”Journal of Manufacturing Systems, vol. 76, pp. 222–233, 2024

    2024

  10. [18]

    Key feature identification for monitoring wafer-to-wafer variation in semiconductor manufacturing,

    S.-K. S. Fan, C.-Y . Hsu, D.-M. Tsai, M. C. Chou, C.-H. Jen, and J.-H. Tsou, “Key feature identification for monitoring wafer-to-wafer variation in semiconductor manufacturing,”IEEE Transactions on Automation Science and Engineering, vol. 19, no. 3, pp. 1530–1541, 2022

    2022

  11. [19]

    Improved yield prediction and failure analysis in semiconductor manufacturing with xgboost and shapley additive explanations models,

    S. Wang and Y . Chen, “Improved yield prediction and failure analysis in semiconductor manufacturing with xgboost and shapley additive explanations models,” in2024 IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits (IPFA), pp. 01–08, IEEE, 2024

    2024

  12. [20]

    A novel approach to reducing testing costs and minimizing defect escapes using dynamic neighborhood range and shapley values,

    T. Ni, W. Rui, C. Zhuo, Y . Li, X. Wen, and M. Nie, “A novel approach to reducing testing costs and minimizing defect escapes using dynamic neighborhood range and shapley values,”ACM Transactions on Design Automation of Electronic Systems, 2025

    2025

  13. [21]

    Efficient estimation of word representations in vector space,

    T. Mikolov, “Efficient estimation of word representations in vector space,”arXiv preprint arXiv:1301.3781, 2013

    2013 arXiv

  14. [22]

    Attention is all you need,

    A. Vaswani, “Attention is all you need,”Advances in Neural Information Processing Systems, 2017

    2017

  15. [23]

    Data-driven optimization of accessory combinations for final testing processes in semiconductor manufacturing,

    S.-K. S. Fan, W.-K. Lin, and C.-H. Jen, “Data-driven optimization of accessory combinations for final testing processes in semiconductor manufacturing,”Journal of Manufacturing Systems, vol. 63, pp. 275– 287, 2022

    2022

  16. [24]

    Graph representation and embedding for semiconductor manufacturing fab states,

    B. Schulz, C. Jacobi, A. Gisbrecht, A. Evangelos, C. W. Chan, and B. P. Gan, “Graph representation and embedding for semiconductor manufacturing fab states,” in2022 Winter Simulation Conference (WSC), pp. 3382–3393, IEEE, 2022

    2022

  17. [25]

    Explain- able AI: A brief survey on history, research areas, approaches and challenges,

    F. Xu, H. Uszkoreit, Y . Du, W. Fan, D. Zhao, and J. Zhu, “Explain- able AI: A brief survey on history, research areas, approaches and challenges,” inProceedings of the 8th cCF international conference on Natural language processing and Chinese computing, part II 8, pp. 563– 574, 2019

    2019

  18. [26]

    A unified approach to interpreting model predictions,

    S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” inProceedings of the 31st International Conference on Neural Information Processing Systems, p. 4768–4777, 2017

    2017

  19. [27]

    ‘Why Should I Trust You?’: Explaining the predictions of any classifier,

    M. T. Ribeiro, S. Singh, and C. Guestrin, “‘Why Should I Trust You?’: Explaining the predictions of any classifier,” inProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD ’16, (New York, NY , USA), p. 1135–1144, 2016

    2016

  20. [28]

    Grad-cam: visual explanations from deep networks via gradient-based localization,

    R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-cam: visual explanations from deep networks via gradient-based localization,”International journal of computer vision, vol. 128, pp. 336–359, 2020

    2020

  21. [29]

    Xai for transformers: Better explanations through conservative propa- gation,

    A. Ali, T. Schnake, O. Eberle, G. Montavon, K.-R. M ¨uller, and L. Wolf, “Xai for transformers: Better explanations through conservative propa- gation,” inInternational Conference on Machine Learning, pp. 435–451, PMLR, 2022

    2022

  22. [30]

    Explainable artificial intelligence (XAI) on timeseries data: A survey,

    T. Rojat, R. Puget, D. Filliat, J. Del Ser, R. Gelin, and N. D ´ıaz- Rodr´ıguez, “Explainable artificial intelligence (XAI) on timeseries data: A survey,”arXiv preprint arXiv:2104.00950, 2021

    2021 arXiv

  23. [31]

    Molnar,Interpretable machine learning

    C. Molnar,Interpretable machine learning. Lulu.com, 2020

    2020

  24. [32]

    Generative perturbation analysis for probabilis- tic black-box anomaly attribution,

    T. Id ´e and N. Abe, “Generative perturbation analysis for probabilis- tic black-box anomaly attribution,” inProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pp. 845–856, 2023

    2023

  25. [33]

    Text classification using string kernels,

    H. Lodhi, C. Saunders, J. Shawe-Taylor, N. Cristianini, and C. Watkins, “Text classification using string kernels,”Journal of machine learning research, vol. 2, no. Feb, pp. 419–444, 2002

    2002

  26. [34]

    Shawe-Taylor and N

    J. Shawe-Taylor and N. Cristianini,Kernel methods for pattern analysis. Cambridge university press, 2004

    2004

  27. [35]

    Causal inference using potential outcomes: Design, modeling, decisions,

    D. B. Rubin, “Causal inference using potential outcomes: Design, modeling, decisions,”Journal of the American statistical Association, vol. 100, no. 469, pp. 322–331, 2005

    2005

  28. [36]

    Visualizing data usingt-SNE.,

    L. Van der Maaten and G. Hinton, “Visualizing data usingt-SNE.,” Journal of machine learning research, vol. 9, no. 11, 2008

    2008

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.