NSR-Boost: A Neuro-Symbolic Residual Boosting Framework for Industrial Legacy Models

2026-01-15

AI Summary

This work addresses the high retraining costs and systemic risks associated with upgrading legacy industrial models—particularly those based on Gradient Boosted Decision Trees (GBDT)—in high-concurrency production environments. The authors propose a non-intrusive neuro-symbolic residual enhancement framework that first identifies “hard regions” where the legacy model fails through residual analysis. Large language models are then leveraged to generate interpretable symbolic rules as expert modules, whose parameters are fine-tuned via Bayesian optimization. A lightweight dynamic aggregator subsequently fuses the outputs of the original model and the expert modules. Crucially, this approach requires no modification or retraining of the legacy system, substantially reducing deployment risk. The method outperforms state-of-the-art approaches across six public benchmarks and a private financial risk-control dataset, and has been successfully deployed in Qfin Holdings’ core risk-management system, yielding significant online performance gains.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Cognitive Modeling & Cognitive Systems: Symbolic Representations
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Although the Gradient Boosted Decision Trees (GBDTs) dominate industrial tabular applications, upgrading legacy models in high-concurrency production environments still faces prohibitive retraining costs and systemic risks. To address this problem, we present NSR-Boost, a neuro-symbolic residual boosting framework designed specifically for industrial scenarios. Its core advantage lies in being"non-intrusive". It treats the legacy model as a frozen model and performs targeted repairs on"hard regions"where predictions fail. The framework comprises three key stages: First, finding hard regions through residuals, then generating interpretable experts by generating symbolic code structures using Large Language Model (LLM) and fine-tuning parameters using Bayesian optimization, and finally dynamically integrating experts with legacy model output through a lightweight aggregator. Experimental results demonstrate that the framework not only significantly outperforms state-of-the-art (SOTA) baselines across six public datasets and one private dataset. More importantly, we report the successful deployment of NSR-Boost within the core financial risk control system of Qfin Holdings, where empirical results on real-world online traffic exhibit superior performance improvements and a significant reduction in the bad rate. In conclusion, it effectively captures long-tail risks missed by traditional models and offers a safe, low-cost evolutionary paradigm for industry.

Problem

• legacy models
• model upgrading
• industrial deployment
• retraining cost
• systemic risk

Innovation

• Neuro-Symbolic
• Residual Boosting
• Legacy Model Upgrade
• Non-intrusive
• Interpretable Experts