Pith. sign in

REVIEW 5 major objections 5 minor 48 references

This paper claims that regional data trading platforms causally strengthen firm supply chain resilience in all five measured dimensions, and that a machine-learning SCM-FSCM stack cuts SME financing costs by 18–22%.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

A DML regression on roughly 40,000 firm-year observations reports positive effects of data element marketization on five supply chain resilience indicators, but the abstract's promised operational improvements are not delivered.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection The paper asks a relevant question about data marketization and supply chain resilience, but the treatment variable is defined inconsistently, the equations are garbled, and the abstract's headline numbers never appear in the body. the 5 major comments →

arxiv 2509.03673 v1 pith:QUS5B5W4 submitted 2025-09-03 cs.LG

A Machine Learning-Based Study on the Synergistic Optimization of Supply Chain Management and Financial Supply Chains from an Economic Perspective

Hang Wang , Huijie Tang , Ningai Leng , Zhoufan Yu This is my paper
classification cs.LG
keywords Supply Chain Management (SCM)Financial Supply Chain Management (FSCM)Machine LearningData element marketizationSupply chain resilienceDouble/Debiased Machine Learning (DML)Causal inference
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 claims that the marketization of data as an economic input — measured by the opening of regional data trading platforms — causally strengthens firm supply chain resilience. Using Double/Debiased Machine Learning on a large firm-year panel, the authors estimate that platform presence improves all five of their resilience dimensions: cooperation stability (significant at 10%), and concentration risk, demand forecasting accuracy, adaptation efficiency, and recovery capability (each at 1%). The same study builds a machine-learning supply-chain and financial-supply-chain model — LSTM forecasting, XGBoost credit scoring, dynamic pledge financing — and reports that in a 20-core-enterprise, 100-supplier simulation it raises inventory turnover by 30%, cuts SME financing costs 18–22%, and holds order fulfillment above 95%. A sympathetic reader would care because the paper offers a causal route from data policy to measurable operational and financing gains, not just a correlation.

Core claim

On its own terms, the paper's central claim is that the marketization of data elements (Mde) causally improves enterprise supply chain resilience. In a partially linear Double Machine Learning model with year and firm fixed effects, the benchmark regressions report significant positive coefficients on Mde for all five resilience dimensions: cooperation stability (SCR1) at the 10% level, and risk resistance, demand forecasting accuracy, adaptation efficiency, and recovery capability (SCR2–SCR5) at the 1% level. The effect is said to run through three channels — technological innovation, transaction-cost reduction, and factor digitalization. Robustness tests (sample exclusion, policy-interfere

What carries the argument

The load-bearing object is the Double/Debiased Machine Learning estimator built on the partially linear model SCR_it+1 = θ0·Mde_it + g0(X_it) + U_it. DML residualizes both the treatment Mde and the outcome SCR against the high-dimensional controls X using cross-validated machine learning (random forest in the benchmark, a neural network as robustness check), then estimates the causal coefficient θ0 from the orthogonalized variation — the step that removes the regularization bias that would otherwise let the nuisance estimate contaminate the target. Around it sit two constructed indices: the five-dimension supply chain resilience composite, and the continuous Data Element Marketization index

Load-bearing premise

The estimate is causal only if, after accounting for the observed controls, the opening of a regional data trading platform is uncorrelated with unobserved factors that also drive supply chain resilience (the conditional-exogeneity condition E[U|X, Mde] = 0), and only if the random-forest-weighted index (0.42/0.35/0.23) genuinely measures data marketization rather than general regional digital development.

What would settle it

Re-estimate the DML model with leads of Mde: if supply chain resilience measurably rises one or two years before a data trading platform opens, the exclusion restriction fails and the causal reading collapses. A second check: re-weight the Mde index with equal weights (or principal components) instead of the random-forest importances; if the benchmark coefficients flip or vanish, the measurement model carries the result. A third: estimate the platform effect with a matched difference-in-differences design and compare magnitudes.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • If the causal estimate holds, a regional data trading platform is a direct policy lever: opening one raises firm supply chain resilience in all five measured dimensions, with the largest benchmark coefficients on risk resistance (0.081) and recovery capability (0.123).
  • Because the coefficients stay significant when random forest is swapped for a neural network and when the cross-validation split changes, the effect is not pinned to one machine-learning implementation — a robustness claim the paper explicitly makes.
  • The model stack converts inventory into working capital: LSTM demand forecasting (error ≤ 8%) and XGBoost credit assessment (accuracy ≥ 90%) feed dynamic pledge financing, which the simulation credits with cutting SME financing costs by 18–22%.
  • The validated H2a–H2c channels imply policy leverage sits in data transaction standards and intelligent platform construction rather than only in firm-level AI adoption — the paper's stated practical takeaway.

Where Pith is reading between the lines

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

  • The paper does not run a pre-trend test: if resilience already rises in the years before a platform opens, the exclusion restriction fails and the causal reading collapses; plotting lead coefficients would settle it.
  • Three load-bearing precedents are cited in text but do not appear in the reference list — Zhou & Wang (2024, Section III.A), Reggiani (2013, Section II.A), and Lin Yue et al. (2023, Section III.B) — leaving the identification template and its measurement precedents untraceable.
  • The Mde index treats random-forest feature importances (0.42/0.35/0.23) as measurement truth; an equal-weight or principal-component re-weighting would show whether the causal coefficients survive the choice of weights.
  • The causal panel analysis and the 20-core/100-supplier simulation are never connected; testing in real firm data whether platform regions show the modeled LSTM-forecast and XGBoost-credit gains — and whether the gains extend beyond the manufacturing sector, the paper's own stated scope limit — is the natural next step.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes a machine-learning-enhanced collaborative model of supply chain management and financial supply chain management, and claims to validate, using Double/Debiased Machine Learning (DML), that the marketization of data elements causally improves five dimensions of firm-level supply chain resilience (SCR1-SCR5). The empirical core is a set of regressions in Tables 1-5, with robustness checks, and the conclusion asserts that Hypotheses H1 and H2a-H2c are supported. The abstract also reports large operational gains (30% inventory turnover increase, 18%-22% financing cost reduction, order fulfillment above 95%) from a verification exercise with 20 core and 100 supporting enterprises.

Significance. If the causal estimates were valid, the paper would offer a useful application of DML to an important policy question—whether regional data trading platforms strengthen firm supply chain resilience—and would contribute a multi-dimensional SCR measurement framework. The attempt to combine economic theory, causal inference, and machine learning is timely. However, the current manuscript does not allow the reader to verify the central claims: the treatment variable is inconsistently defined, the DML equations are garbled, the mediation results are asserted rather than reported, and the headline quantitative results in the abstract are absent from the empirical sections. The paper therefore cannot currently support its conclusions.

major comments (5)
  1. [§III.A vs. §III.B.2] The treatment variable is defined inconsistently. Equations (1)-(4) define Med_it as a binary policy dummy: '1 if the region of firm i established a data trading platform in year t, otherwise 0.' In contrast, §III.B.2 defines the explanatory variable as a continuous 'Data Element Marketization Index' built from three indicators with random-forest feature-importance weights (0.42, 0.35, 0.23) and Z-score standardization. Tables 1-5 report coefficients labeled 'Mde' without specifying which definition is used. If the continuous index is used, the binary-policy-shock interpretation of equations (1)-(4) is false; if the dummy is used, the measurement model in §III.B.2 is not the variable being estimated. Either way, the reported coefficients do not correspond to a well-defined causal estimand, undermining the central claim.
  2. [§III.A, Eqs. (6)-(9)] The DML estimator equations are garbled and internally incoherent. Equation (6) contains malformed terms such as '∑_i∈I M Med_it' and 'g0̂(Xi)' without clear definitions. The text states: 'as b → ∞, it follows that n(θ̂0 − θ0) →p ∞,' which is the opposite of the consistency DML is supposed to establish. The final estimating equation—the second display labeled (8)—mixes V, V̂, and Mde in a way that does not match the preceding derivation. Equation (9) also has unresolved notation. These errors prevent the reader from verifying that the method implemented in Tables 1-5 is actually DML, and they cast doubt on the validity of the reported significance levels.
  3. [§V vs. §IV] The conclusion states that 'the multi-channel transmission mechanism also validates Hypotheses H2a-H2c' (technological innovation, transaction cost reduction, and factor digitalization), but no mediation analysis is presented anywhere in §IV. There are no tables or figures reporting the effects of Mde on Tech_inno, Trans_cost, or Fin_sync, nor any test of the indirect paths. The only reported results are benchmark and robustness regressions of Mde on SCR1-SCR5. The mediation hypotheses are therefore asserted, not supported, and this is load-bearing for the paper's mechanism story.
  4. [Abstract and §IV (all)] The abstract's headline quantitative claims—30% increase in inventory turnover, 18%-22% decrease in SME financing costs, order fulfillment rate above 95%, demand forecasting error ≤ 8%, credit assessment accuracy ≥ 90%—are never derived or reported in the empirical sections. The verification exercise involving '20 core and 100 supporting enterprises' is not described: no sample description, no simulation setup, no table of results. These metrics are central to the paper's stated contribution, and their absence makes the abstract's claims unverifiable.
  5. [§III.B.2 and §III.B.1 (SCR3)] There is a circularity problem in the construction of key variables. SCR3 is defined as the reciprocal of the absolute deviation between the LSTM prediction and actual sales, but the paper does not state whether the LSTM is trained on the same sample used in the regression. If in-sample, SCR3 measures model fit, not out-of-sample demand forecasting accuracy, and using it as a dependent variable in a causal regression is circular. Similarly, the Mde index weights are obtained from a random forest regression on feature importance, but the target variable for that random forest is unspecified; if it is the same SCR outcome used in Tables 1-5, the index is endogenous by construction. The weights (0.42, 0.35, 0.23) are also treated as known rather than estimated with uncertainty. These issues affect the interpretation of all Mde coefficients.
minor comments (5)
  1. [Table 1 note] The significance-star note reads '***, *, * indicate significance at 1%, 5%, 10% levels,' which appears garbled; presumably it should be '***, **, *'.
  2. [§III.A equation numbering] Two equations are both labeled (8), and equation (5) uses X_{1t}, ..., X_{pt} while the text uses X_it. This makes the notation hard to follow.
  3. [Figures 1 and 2] The paper includes Fig. 1 (CNN) and Fig. 2 (RNN) with no substantive description or architectural details; they do not appear to be referenced in the analysis.
  4. [References] A large fraction of the reference list (e.g., [6]-[41]) is unrelated to the paper's topic, including papers on SAR-ATR, golf swing analysis, and museum souvenirs, while key claims such as the DML approach are only attributed to a single source. This reduces the paper's scholarly grounding.
  5. [Tables 2-5] The sample sizes vary across columns within each table (e.g., Table 2: 29,445 to 33,195) without explanation. The coefficients also change substantially across robustness specifications (e.g., SCR1: 0.038 in Table 1 to 0.012 in Table 2 to 0.018 in Table 3), but the text does not discuss these fluctuations.

Circularity Check

2 steps flagged

Moderate circularity: the Mde treatment is a random-forest-weighted index fitted on the analysis data and then regressed on SCR outcomes, while the DML model's identifying assumptions refer to a different binary policy variable.

specific steps
  1. fitted input called prediction [§III.B.2 and §IV.A (Table 1)]
    "Instead of the traditional dummy variable, this paper uses a continuous variable "Data Element Marketization Index" for measurement ... Weight Calculation: Use random forest regression algorithm to score the feature importance of each indicator, with weights of 0.42, 0.35, and 0.23 respectively; Standardization: Convert raw data into an index value in the range [0,1] through Z-score standardization."

    The treatment variable used in Tables 1-5 is not the exogenous policy dummy for which the DML estimator in Eqs. (1)-(8) was built. It is a composite index whose weights are random-forest feature-importance scores estimated from the data. Feature importance is a supervised quantity; the paper never specifies the target used for the scoring. In a study whose outcome is supply chain resilience, the weights can be (and as written are) fitted to the same SCR outcomes that the regression later 'explains.' The coefficient θ0 therefore partly measures the index's own construction rather than an exogenous effect of data marketization.

  2. self definitional [§III.A vs §III.B.2]
    "Medit (policy variable): 1 if the region of firm i established a data trading platform in year t, otherwise 0 ... Instead of the traditional dummy variable, this paper uses a continuous variable "Data Element Marketization Index" for measurement."

    The causal estimand is defined twice: the DML model in §III.A treats Mde as a binary policy shock, while §III.B.2 replaces it with a continuous random-forest-weighted index. The tables report a single coefficient labeled Mde without saying which definition was used. If the continuous index is used, the exogeneity assumption E[U|X,Med]=0 in Eq. (2) applies to a fitted composite rather than to a policy shock; if the dummy is used, the §III.B.2 measurement model is irrelevant. Either way, the reported 'effect' is not attached to a single well-defined treatment, so the conclusion that marketization improves all five SCR dimensions conflates two different objects.

full rationale

The central causal claim is not self-contained against external benchmarks. The main circularity risk is the construction of Mde: the paper fits random-forest feature-importance weights to data and then uses the resulting index as the treatment in a DML regression predicting SCR outcomes from the same data. Because no independent target for the feature-importance scoring is given, the significant coefficients in Tables 1-5 are at least partly artifacts of the index construction. The additional binary-vs-continuous equivocation means Eqs. (1)-(8) do not apply to the variable actually used. Separately, the paper asserts H2a-H2c are validated without reporting mediation regressions, and the 'Zhou & Wang (2024)' citation on which the exogenous-shock assumption rests is missing from the reference list; these are correctness/transparency problems rather than circularity. SCR3 also embeds an LSTM forecast error without specifying a train/test split, so some of the resilience outcome is a fitted quantity. The abstract's SCM-FSCM simulation results (30% inventory turnover, 18-22% financing cost reduction) are not tied to the DML analysis and do not themselves create circularity. Overall, partial circularity: the 'prediction' of Mde's effect is not independent of how Mde was constructed.

Axiom & Free-Parameter Ledger

4 free parameters · 5 axioms · 0 invented entities

The analysis rests on a treatment variable whose construction is data-dependent, on a standard but untestable exogeneity assumption, and on several mediator variables built from arbitrary weights and keyword sets. The abstract's operational results depend on an unshown 120-enterprise validation.

free parameters (4)
  • Mde index weights = 0.42, 0.35, 0.23
    Random forest feature importance scores fitted on the same data used to estimate the treatment effect; used to build the continuous data element marketization index.
  • Top-3 partner threshold for SCR1/SCR2
    The stability and concentration indicators are defined by the proportion of the top 3 partners; the threshold is chosen by the authors without justification.
  • Patent score weights for Tech_inno = invention x2, utility model x1
    Arbitrary weighting of patent types in the mediating variable for technological innovation.
  • Text-mining keyword set for Trans_cost
    The transaction cost mediator is built from keyword frequencies such as 'coordination cost' and 'information cost', but the keyword set is not listed.
axioms (5)
  • domain assumption Unconfoundedness: E[Uit|Xit, Medit]=0
    Stated in equation (2); requires that all confounders of the data marketization-resilience relationship are captured by the controls.
  • domain assumption Exogeneity of regional data trading platform establishment
    Invoked in Section III A, 'we use regional data trading platforms as proxy exogenous policy shocks'; no evidence rules out anticipation or location selection effects.
  • domain assumption Random forest and LSTM nuisance models are correctly specified
    DML relies on consistent estimates of g0 and m0; the paper provides no tuning details or validation.
  • ad hoc to paper Measurement model for Mde index is valid
    The index combines three indicators with weights from a random forest fitted on the same sample; no external benchmark or robustness to alternative weights is provided.
  • ad hoc to paper SCR3 demand forecast accuracy is computed from a valid out-of-sample LSTM
    SCR3 is defined as 1 minus the LSTM forecast error, but no training/validation split or model details are given.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of A Machine Learning-Based Study on the Synergistic Optimization of Supply Chain Management and Financial Supply Chains from an Economic Perspective." pith.science (2026). https://pith.science/paper/QUS5B5W4

@misc{pith2026250903673,
  author       = {Pith},
  title        = {Pith review of: A Machine Learning-Based Study on the Synergistic Optimization of Supply Chain Management and Financial Supply Chains from an Economic Perspective},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QUS5B5W4}},
  note         = {Machine review of arXiv:2509.03673}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Based on economic theories and integrated with machine learning technology, this study explores a collaborative Supply Chain Management and Financial Supply Chain Management (SCM - FSCM) model to solve issues like efficiency loss, financing constraints, and risk transmission. We combine Transaction Cost and Information Asymmetry theories and use algorithms such as random forests to process multi-dimensional data and build a data-driven, three-dimensional (cost-efficiency-risk) analysis framework. We then apply an FSCM model of "core enterprise credit empowerment plus dynamic pledge financing." We use Long Short-Term Memory (LSTM) networks for demand forecasting and clustering/regression algorithms for benefit allocation. The study also combines Game Theory and reinforcement learning to optimize the inventory-procurement mechanism and uses eXtreme Gradient Boosting (XGBoost) for credit assessment to enable rapid monetization of inventory. Verified with 20 core and 100 supporting enterprises, the results show a 30\% increase in inventory turnover, an 18\%-22\% decrease in SME financing costs, a stable order fulfillment rate above 95\%, and excellent model performance (demand forecasting error <= 8\%, credit assessment accuracy >= 90\%). This SCM-FSCM model effectively reduces operating costs, alleviates financing constraints, and supports high-quality supply chain development.

Figures

Figures reproduced from arXiv: 2509.03673 by Hang Wang, Huijie Tang, Ningai Leng, Zhoufan Yu.

Figure 1
Figure 1. Figure 1: CNN (Convolutional Neural Network) Model From an external enterprise perspective, deep learning underpins the collaborative resilience of the supply chain network: On one hand, relying on federated deep learning, data from multiple entities is aggregated to train models under the premise of protecting privacy, improving the accuracy of decisions such as credit evaluation. On the other hand, professional in… view at source ↗
Figure 2
Figure 2. Figure 2: RNN (Recurrent Neural Network) Model B. Mechanism of the Multi-Channel Impact of Data Element Marketization on Enterprise Supply Chain Resilience 1) Enterprise Technological Innovation: Core Driver of Algorithm Iteration and Model Optimization The marketization of data elements provides "data fuel" and "algorithm engine" for enterprise technological innovation, and strengthens supply chain resilience in bo… view at source ↗
Figure 3
Figure 3. Figure 3: Analysis framework and influence path III. DESIGN OF THE STUDY A. Model-building To verify the causal relationship between digital element marketization and enterprise supply chain resilience, we use the Double Machine Learning (DML) model for empirical testing. Ordinary multiple linear regression has three flaws: (1) Exponential decline in training sample density with increasing covariate dimensions cause… view at source ↗

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

48 extracted references · 41 canonical work pages · 2 internal anchors

  1. [1]

    combined with RNN (Recurrent Neural Network, shown in Figure 2) to generate user portraits, and deeply optimize genetic algorithms for inventory allocation, achieving end- to-end intelligent linkage.Hypothesis H1: The marketization of data elements can enhance the resilience of enterprise supply chains through deep learning -driven data perception, algori...

    work page

  2. [2]

    data fuel

    Enterprise Technological Innovation: Core Driver of Algorithm Iteration and Model Optimization The marketization of data elements provides "data fuel" and "algorithm engine" for enterprise technological innovation, and strengthens supply chain resilience in both directions through expansion of model innovation breadth and improvement of algorithm optimiza...

    work page

  3. [3]

    usable but invisible

    Supply Chain Transaction Costs: Cost Reduction Paths through Data Governance and Intelligent Collaboration Relying on data governance technologies and intelligent interaction systems, the marketization of data elements fundamentally reduces supply chain transaction costs and indirectly improves resilience. Firstly, data compliance and privacy computing te...

    work page

  4. [4]

    human-machine collaboration

    Digitalization of Traditional Production Factors: Multiplier Effect of Factor Empowerment and Model Integration Through factor digital modeling and intelligent collaboration algorithms, the marketization of data elements activates the digital multiplier effect of traditional production factors and reconstructs the operation logic of the supply chain.\ At ...

    work page 2024

  5. [5]

    risk resistance - dynamic adjustment - collaborative recovery

    Dependent Variable: Enterprise Supply Chain Resilience (SCR) Existing studies mostly measure supply chain resilience from a single dimension, which fails to fully reflect the complete chain of "risk resistance - dynamic adjustment - collaborative recovery". Combining the advantages of machine learning algorithms in fusing multi -dimensional features, this...

    work page 2023

  6. [6]

    Data Element Marketization Index

    Explanatory Variable: Marketization of Data Elements (Mde) Instead of the traditional dummy variable, this paper uses a continuous variable "Data Element Marketization Index" for measurement, which is more suitable for machine learning's processing needs of continuous features. The index is constructed through the following three steps: Basic Indicator La...

    work page

  7. [7]

    Broadband China

    show data element marketization significantly boosts supply chain resilience across dimensions: 10% significance for stability (SCR1) and 1% significance for risk resistance (SCR2), predictive capability (SCR3), adaptation capability (SCR4), and recovery capability (SCR5). Hypothesis H1 is validated. TABLE I. BENCHMARK REGRESSION RESULTS (1) SCR1 (2) SCR2...

    work page

  8. [8]

    Integrating deep learning and reinforcement learning for enhanced financial risk forecasting in supply chain management[J]

    Cui Y, Yao F. Integrating deep learning and reinforcement learning for enhanced financial risk forecasting in supply chain management[J]. Journal of the Knowledge Economy, 2024, 15(4): 20091-20110

    work page 2024

  9. [9]

    Sustainable optimization in supply chain management using machine learning[J]

    Zhu J, Wu Y, Liu Z, et al. Sustainable optimization in supply chain management using machine learning[J]. International Journal of Management Science Research, 2025, 8(1): 1-8

    work page 2025

  10. [10]

    Optimizing green supply chain circular economy in smart cities with integrated machine learning technology[J]

    Liu T, Guan X, Wang Z, et al. Optimizing green supply chain circular economy in smart cities with integrated machine learning technology[J]. Heliyon, 2024, 10(9)

    work page 2024

  11. [11]

    Machine learning in supply chain management: systematic literature review and future research agenda[J]

    Vlachos I, Reddy P G. Machine learning in supply chain management: systematic literature review and future research agenda[J]. International Journal of Production Research, 2025: 1-30

    work page 2025

  12. [12]

    Machine Learning Models for Optimizing Supply Chain Management[C]//2024 8th International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT)

    Alkubaisi S, Fenjan A, Al- Sharify T A, et al. Machine Learning Models for Optimizing Supply Chain Management[C]//2024 8th International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT). IEEE, 2024: 1-9

    work page 2024

  13. [13]

    Xiong, X., Zhang, X., Jiang, W., Liu, T., Liu, Y., & Liu, L. (2024). Lightweight dual -stream SAR -ATR framework based on an attention mechanism -guided heterogeneous graph network. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 1–22

    work page 2024

  14. [14]

    Wang, J., Zhang, Z., He, Y., Song, Y., Shi, T., Li, Y., ... & He, L. (2024). Enhancing Code LLMs with Reinforcement Learning in Code Generation. arXiv preprint arXiv:2412.20367

    Pith/arXiv arXiv 2024

  15. [15]

    Wang, B. (2025). Empirical Evaluation of Large Language Models for Asset -Return Prediction. Academic Journal of Sociology and Management, 3(4), 18–25

    work page 2025

  16. [16]

    (2024, October)

    Kang, Y., Song, Y., & Huang, S. (2024, October). Tie memories to e- souvenirs: Personalized souvenirs with augmented reality for interactive learning in the museum. Preprints

    work page 2024

  17. [17]

    (2024, December)

    Feng, C., Bačić, B., & Li, W. (2024, December). Sca - lstm: A deep learning approach to golf swing analysis and performance enhancement. In International Conference on Neural Information Processing (pp. 72- 86). Singapore: Springer Nature Singapore

    work page 2024

  18. [18]

    Zhang, L., Wang, L., Huang, Y., & Chen, H. (2019). Segmentation of Thoracic Organs at Risk in CT Images Combining Coarse and Fine Network. SegTHOR@ ISBI, 11(16), 2-4

    2019

  19. [19]

    (2022, January 20– 22)

    Liu, Z. (2022, January 20– 22). Stock volatility prediction using LightGBM based algorithm. In 2022 International Conference on Big Data, Information and Computer Network (BDICN) (pp. 283–286). IEEE

    work page 2022

  20. [20]

    Zheng, Z., Liu, K., & Zhu, X. (2025). Machine learning- based prediction of metal-organic framework materials: A comparative analysis of multiple models. arXiv preprint arXiv:2507.04493

    Pith/arXiv arXiv 2025

  21. [21]

    Huang, S., Song, Y., Kang, Y., Yu, C., et al. (2024). AR overlay: Training image pose estimation on curved surface in a synthetic way. In CS & IT Conference Proceedings (Vol. 14, No. 17). CS & IT Conference Proceedings

    work page 2024

  22. [22]

    Zhou, Y., Shen, J., & Cheng, Y. (2025). Weak to strong generalization for large language models with multi- capabilities. In The Thirteenth International Conference on Learning Representations

    work page 2025

  23. [23]

    Li, K., Liu, L., Chen, J., Yu, D., Zhou, X., Li, M., ... & Li, Z. (2024, November). Research on reinforcement learning based warehouse robot navigation algorithm in complex warehouse layout. In 2024 6th International Conference on Artificial Intelligence and Computer Applications (ICAICA) (pp. 296-301). IEEE

    work page 2024

  24. [24]

    Wang, H., Sun, W., & Liu, Y. (2022). Prioritizing autism risk genes using personalized graphical models estimated from single- cell rna-seq data. Journal of the American Statistical Association, 117(537), 38-51

    work page 2022

  25. [25]

    Li, J., & Zhou, Y. (2025). Bideeplab: An improved lightweight multi-scale feature fusion deeplab algorithm for facial recognition on mobile devices. Computer Simulation in Application, 3(1), 57–65

    work page 2025

  26. [26]

    Li, X., & Fang, L. (2025). The Impact of Culturally Adaptive Feedback in Information -Interaction Systems on Language -Learning Motivation and Outcomes. Journal of Educational Theory, 2(1), 1–7

    work page 2025

  27. [27]

    & Wang, X

    He, Y., Wang, J., Li, K., Wang, Y., Sun, L., Yin, J., ... & Wang, X. (2025). Enhancing Intent Understanding for Ambiguous Prompts through Human- Machine Co - Adaptation. arXiv preprint arXiv:2501.15167

    arXiv 2025

  28. [28]

    Research and Design of Blockchain-based Propagation Algorithm for IP Tags Using Local Random Walk in Big Data Marketing

    Yu Jiang, Tianzuo Zhang, Huanyu Liu. Research and Design of Blockchain-based Propagation Algorithm for IP Tags Using Local Random Walk in Big Data Marketing. TechRxiv. June 06, 2025. DOI: 10.36227/techrxiv.174918179.99038154/v1

    work page arXiv 2025

  29. [29]

    Ning, Z., Zeng, H., & Tian, Z. (2025). Research on data- driven energy efficiency optimisation algorithm for air compressors. In Third International Conference on Advanced Materials and Equipment Manufacturing (AMEM 2024) (Vol. 13691, pp. 1068–1075). SPIE

    work page 2025

  30. [30]

    Zhou, Z. (2025). Research on the Application of Intelligent Robots and Software in Multiple Segmentation Scenarios Based on Machine Learning. Available at SSRN 5236930

    work page 2025

  31. [31]

    & Zhu, A

    Zhao, P., Liu, X., Su, X., Wu, D., Li, Z., Kang, K., ... & Zhu, A. (2025). Probabilistic Contingent Planning Based on Hierarchical Task Network for High -Quality Plans. Algorithms, 18(4), 214

    work page 2025

  32. [32]

    Liu, Z. (2025). Human -AI Co-Creation: A Framework for Collaborative Design in Intelligent Systems. arXiv:2507.17774

    Pith/arXiv arXiv 2025

  33. [33]

    Wang, J., Ding, W., & Zhu, X. (2025). Financial analysis: Intelligent financial data analysis system based on LLM-RAG. arXiv preprint arXiv:2504.06279

    Pith/arXiv arXiv 2025

  34. [34]

    Zhou, Z., Guo, Y., Guo, L., Hong, J., & Peng, J. (2025). Research on Robot Process Modeling and Software Engineering Integration Path in Automated Processing of Accounting Information Based on Artificial Intelligence and Big Data. Available at SSRN 5359072

    work page 2025

  35. [35]

    Yu, D., Liu, L., Wu, S., Li, K., Wang, C., Xie, J., ... & Ji, R. (2025, March). Machine learning optimizes the efficiency of picking and packing in automated warehouse robot systems. In 2025 IEEE International Conference on Electronics, Energy Systems and Power Engineering (EESPE) (pp. 1325-1332). IEEE

    work page 2025

  36. [36]

    Bačić, B., Vasile, C., Feng, C., & Ciucă, M. G. (2024). Towards nation -wide analytical healthcare infrastructures: A privacy -preserving augmented knee rehabilitation case study. arXiv preprint arXiv:2412.20733

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  37. [37]

    T., Pang, P

    Niu, T., Liu, T., Luo, Y. T., Pang, P. C.-I., Huang, S., & Xiang, A. (2025). Decoding student cognitive abilities: A comparative study of explainable AI algorithms in educational data mining. Scientific Reports, 15(1),

    work page 2025

  38. [38]

    Yu, Z., Tang, H., & Leng, N. (2025). Spillover Effects of Carbon Emission Trading on the Green Supply-Chain Finance Network. Journal of Economic Theory and Business Management, 2(3), 1–5

    work page 2025

  39. [39]

    Bačić, B., Feng, C., & Li, W. (2024). Jy61 imu sensor external validity: A framework for advanced pedometer algorithm personalisation. ISBS Proceedings Archive, 42(1), 60

    work page 2024

  40. [40]

    Zhou, Y., Zhang, J., Chen, G., Shen, J., & Cheng, Y. (2024). Less is more: Vision representation compression for efficient video generation with large language models

    work page 2024

  41. [41]

    (2025, August)

    Huang, S., Kang, Y., Shen, G., & Song, Y. (2025, August). AI -augmented context- aware generative pipelines for 3D content. Preprints. https://doi.org/10.20944/preprints202508.0195.v1

    work page arXiv 2025

  42. [42]

    Wang, P., Wang, H, Li, Q., Shen, D., & Liu, Y. (2024). Joint and individual component regression. Journal of Computational and Graphical Statistics, 33(3), 763-773

    work page 2024

  43. [43]

    Warehouse Robot Task Scheduling Based on Reinforcement Learning to Maximize Operational Efficiency

    Siye Wu, Lei Fu, Runmian Chang, et al. Warehouse Robot Task Scheduling Based on Reinforcement Learning to Maximize Operational Efficiency. TechRxiv. April 18, 2025.DOI: 10.36227/techrxiv.174495431.17315991/v1

    work page arXiv 2025

  44. [44]

    Zhong, Z., Yang, S., & Becigneul, G. (2021). Environment and speaker-related emotion recognition in conversations. In The 2nd International Conference on Computing and Data Science (pp. 1–6)

    work page 2021

  45. [45]

    Yang, S., Zhao, Y., & Gao, H. (2024). Using large language models in real estate transactions: A few- shot learning approach. OSF Preprints

    work page 2024

  46. [46]

    (2025, May 28)

    Ding, T., Xiang, D., Qi, Y., Yang, Z., Zhao, Z., Sun, T., Feng, P., & Wang, H. (2025, May 28). NeRF -based defect detection. In Proceedings of SPIE—International Conference on Remote Sensing, Mapping, and Image Processing (RSMIP 2025) (Vol. 13650, 136501E). SPIE. https://doi.org/10.1117/12.3067707

    work page doi:10.1117/12.3067707 2025

  47. [47]

    Ding, T., Xiang, D., Sun, T., Qi, Y., & Zhao, Z. (2025). AI-driven prognostics for state of health prediction in Li-ion batteries: A comprehensive analysis with validation. arXiv preprint arXiv:2504.05728

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  48. [48]

    Ding, T., Xiang, D., Rivas, P., & Dong, L. (2025). Neural pruning for 3D scene reconstruction: Efficient NeRF acceleration. arXiv preprint arXiv:2504.00950

    Pith/arXiv arXiv 2025

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.