Yanan Jiao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Yanan Jiao has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Wenjie Ping
Georgetown University, Washington, DC, USA
2 shared publications Huijie Fan
Hunan Agricultural University, Changsha, China
1 shared publication- Huijie Fan
Hunan Agricultural University, Changsha, China
1 shared publication - Xu Han
Renmin University of China, Beijing, China
1 shared publication - Xuguang Zhang
University of Gloucestershire, Cheltenham, U.K.
1 shared publication - Yiwen Liang
SC Johnson Graduate School of Management, Cornell University, Ithaca, NY 14853, USA
1 shared publication
Latest Publications
- Open Access
- 13 November 2026
- Published by Elsevier in SSRN Electronic Journal
Accounting-grounded Explainable Machine Learning for Financial Statement Misstatement Detection: Temporal Validation Using U.S. Public-Company Filings
Machine-learning detectors of financial statement misstatement discriminate well on raw statement items, but their rankings are difficult to justify in accounting terms and are usually evaluated under protocols that leak information across serial misstatements. We present AGEX, an accounting-grounded explainable detector that couples a library of 44 theory-derived risk indicators (accrual anomalies, earnings-cash divergence, revenue and asset quality, cross-statement articulation checks, and incentives) with peer anchoring inside operating-structure segments, theory-signed monotone gradient boosting, and a scalecontext view built on raw line items. The two views are fused by within-year rank averaging and explained with exact TreeSHAP attributions aggregated into indicator families. On 141,463 U.S. firm-years (1991-2014) labeled by SEC Accounting and Auditing Enforcement Releases, we adopt a label-maturity-aware protocol that trains on years up to T-2 and purges serial misstatement cases. Over the test years 2003-2011, AGEX attains a mean AUC of 0.732, compared with 0.720 for a tuned gradient-boosting model on raw items, 0.707 for RUSBoost, 0.668 for the F-score, and 0.551 for the M-score, and it has the smallest cross-year dispersion among trained models in 2003-2008. Leaky evaluation protocols inflate AUC by up to 0.107 and reverse model rankings. Monotone priors make all learned indicator effects consistent with accounting theory (60.3% without constraints) and raise the crossyear stability of explanations from 0.818 to 0.907 at a cost below 0.006 AUC. At a 10% inspection budget, AGEX captures 30.7% of misstated firm-years and at least one misstated year of 42.4% of enforcement cases.
- 13 November 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAccounting-grounded Explainable Machine Learning for Financial Statement Misstatement Detection: Temporal Validation Using U.S. Public-Company Filings
Machine-learning detectors of financial statement misstatement discriminate well on raw statement items, but their rankings are difficult to justify in accounting terms and are usually evaluated under protocols that leak information across serial misstatements. We present AGEX, an accounting-grounded explainable detector that couples a library of 44 theory-derived risk indicators (accrual anomalies, earnings-cash divergence, revenue and asset quality, cross-statement articulation checks, and incentives) with peer anchoring inside operating-structure segments, theory-signed monotone gradient boosting, and a scalecontext view built on raw line items. The two views are fused by within-year rank averaging and explained with exact TreeSHAP attributions aggregated into indicator families. On 141,463 U.S. firm-years (1991-2014) labeled by SEC Accounting and Auditing Enforcement Releases, we adopt a label-maturity-aware protocol that trains on years up to T-2 and purges serial misstatement cases. Over the test years 2003-2011, AGEX attains a mean AUC of 0.732, compared with 0.720 for a tuned gradient-boosting model on raw items, 0.707 for RUSBoost, 0.668 for the F-score, and 0.551 for the M-score, and it has the smallest cross-year dispersion among trained models in 2003-2008. Leaky evaluation protocols inflate AUC by up to 0.107 and reverse model rankings. Monotone priors make all learned indicator effects consistent with accounting theory (60.3% without constraints) and raise the crossyear stability of explanations from 0.818 to 0.907 at a cost below 0.006 AUC. At a 10% inspection budget, AGEX captures 30.7% of misstated firm-years and at least one misstated year of 42.4% of enforcement cases.
- Top Cited PapersOpen Access
- 20 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 80456-80468
Multimodal Fraud Detection in Financial Statements: A Trimodal Attention Network With Contrastive Evidence Chain Construction
Financial statement fraud inflicts enormous economic damage and continues to evade conventional detection methods that rely predominantly on structured accounting ratios. Such approaches discard the rich forensic signals present in narrative disclosures and document formatting—two channels that fraudsters routinely manipulate alongside the numbers themselves. This paper introduces TriAudit, a trimodal deep learning framework that jointly encodes three complementary evidence streams from corporate 10-K filings: (i) structured financial ratios and accrual indicators processed through a residual MLP, (ii) MDA textual semantics captured via a FinBERT-based encoder with cross-chunk attention, and (iii) visual document-layout features extracted using a LayoutLMv2 backbone augmented with a novel Layout Deviation Score that quantifies spatial anomalies against industry-normative templates. These modality-specific representations are integrated through a hierarchical cross-modal attention mechanism—mirroring the auditor’s evidential reasoning from numerical signals through narrative context to document structure—followed by a gated fusion module that adaptively weights each stream based on perfiling reliability. To support auditability, TriAudit incorporates a contrastive evidence-chain module that aligns anomaly scores across modalities into structured, human-interpretable audit trails linking specific financial irregularities, linguistic deviations, and layout inconsistencies. Evaluated on over 6,000 filings from the EDGAR Corpus with fraud labels derived from SEC Accounting and Auditing Enforcement Releases, TriAudit achieves an AUC-ROC of 0.923 and F1 of 0.847, surpassing the strongest bimodal baseline by 5.5 AUC points and the classical Beneish M-score by over 18 points. Ablation experiments confirm that each architectural component—including the layout deviation scoring, hierarchical attention ordering, contrastive alignment, and orthogonality regularization—contributes meaningfully to both predictive accuracy and interpretability.
- 20 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 80456-80468
Top Cited PapersOpen AccessMultimodal Fraud Detection in Financial Statements: A Trimodal Attention Network With Contrastive Evidence Chain Construction
Financial statement fraud inflicts enormous economic damage and continues to evade conventional detection methods that rely predominantly on structured accounting ratios. Such approaches discard the rich forensic signals present in narrative disclosures and document formatting—two channels that fraudsters routinely manipulate alongside the numbers themselves. This paper introduces TriAudit, a trimodal deep learning framework that jointly encodes three complementary evidence streams from corporate 10-K filings: (i) structured financial ratios and accrual indicators processed through a residual MLP, (ii) MDA textual semantics captured via a FinBERT-based encoder with cross-chunk attention, and (iii) visual document-layout features extracted using a LayoutLMv2 backbone augmented with a novel Layout Deviation Score that quantifies spatial anomalies against industry-normative templates. These modality-specific representations are integrated through a hierarchical cross-modal attention mechanism—mirroring the auditor’s evidential reasoning from numerical signals through narrative context to document structure—followed by a gated fusion module that adaptively weights each stream based on perfiling reliability. To support auditability, TriAudit incorporates a contrastive evidence-chain module that aligns anomaly scores across modalities into structured, human-interpretable audit trails linking specific financial irregularities, linguistic deviations, and layout inconsistencies. Evaluated on over 6,000 filings from the EDGAR Corpus with fraud labels derived from SEC Accounting and Auditing Enforcement Releases, TriAudit achieves an AUC-ROC of 0.923 and F1 of 0.847, surpassing the strongest bimodal baseline by 5.5 AUC points and the classical Beneish M-score by over 18 points. Ablation experiments confirm that each architectural component—including the layout deviation scoring, hierarchical attention ordering, contrastive alignment, and orthogonality regularization—contributes meaningfully to both predictive accuracy and interpretability.
- Top Cited PapersOpen Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 1
Adaptive Event-Driven Labeling: A Neuro-Symbolic Multiagent Framework for Causal Inference in Non-Stationary Time Series
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 1
Top Cited PapersOpen AccessAdaptive Event-Driven Labeling: A Neuro-Symbolic Multiagent Framework for Causal Inference in Non-Stationary Time Series