Paper: arXiv 2510.26627

Authors: Dmitry Lesnik, Tobias Schaefer

Abstract

Machine learning models used for high-stakes predictions in domains like credit risk face critical degradation due to concept drift, requiring robust and transparent adaptation mechanisms. We propose an architecture, where a dedicated correction layer is employed to efficiently capture systematic shifts in predictive scores when a model becomes outdated. The key element of this architecture is the design of a correction layer using Probabilistic Rule Models (PRMs) based on Markov Logic Networks, which guarantees intrinsic interpretability through symbolic, auditable rules. This structure transforms the correction layer from a simple scoring mechanism into a powerful diagnostic tool capable of isolating and explaining the fundamental changes in borrower riskiness. We illustrate this diagnostic capability using Fannie Mae mortgage data, demonstrating how the interpretable rules extracted by the correction layer successfully explain the structural impact of the 2008 financial crisis on specific population segments, providing essential insights for portfolio risk management and regulatory compliance.

Complexity vs Empirical Score

  • Math Complexity: 5.5/10
  • Empirical Rigor: 6.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces a novel application of Probabilistic Rule Models based on Markov Logic Networks to construct an interpretable correction layer for concept drift in credit risk. It offers strong theoretical foundations combining statistical mechanics and formal logic, supported by empirical experiments on Fannie Mae mortgage data around the 2008 financial crisis.

Research Flowchart

  flowchart TD
    A["Research Goal: Diagnose & Adapt to Post-Crisis Concept Drift in Credit Risk Models"] --> B["Data Input: Fannie Mae Mortgage Data (Pre/Post 2008 Crisis)"]
    B --> C["Base ML Model: Predict Borrower Risk (Becomes Outdated)"]
    C --> D["Methodology: Add Probabilistic Rule Model (PRM) Correction Layer via Markov Logic Networks"]
    D --> E["Computational Process: Train Symbolic Rules on Score Drift & Evaluate Risk Shifts"]
    E --> F["Key Outcome: Transparent Diagnostics Isolating Structural Changes for Risk & Compliance"]

Cited References (20)

  1. Mortgage-Backed Securities and the Financial Crisis of 2008: A POST MORTEM (2018) — link
  2. Crafting Papers on Machine Learning (2000)
  3. The Need for Biases in Learning Generalizations (1980)
  4. Computational Complexity of Machine Learning (1989)
  5. Machine Learning: An Artificial Intelligence Approach, Vol. I (1983)
  6. Markov logic networks. Mach Learn 62, 107–136. (2006)
  7. An optimized algorithm for association rule mining using FP tree. (2015)
  8. Markov Random Field modeling, inference & learning in computer vision & image understanding: A survey (2013)
and 12 more
  1. Probabilistic Logic (1986)
  2. Beyond Market Baskets: Generalizing Association Rules to Dependence Rules (1998)
  3. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. (2019)
  4. Hinge-loss Markov random fields and probabilistic soft logic (2017)
  5. Prediction of viral symptoms using wearable technology and artificial intelligence: A pilot study in healthcare workers (2021)
  6. Fannie Mae Single - Family Loan Performance Data (2024)
  7. Probabilistic graphical models: principles and techniques. (2009)
  8. Practical Methods of Optimization. (1987)
  9. Credit Intelligence & Modelling: Many Paths through the Forest of Credit Rating and Scoring (2021)
  10. Probabilistic reasoning in intelligent systems: Networks of plausible inference. (1988)
  11. Evidence-Based Clustering for Scalable Inference in Markov Logic (2014)
  12. Mining association rules between sets of items in large databases (1993)

0 of 20 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.