Judicial Sentencing Prediction Based on Hybrid Models and Two-Stage Learning Algorithms
Ruifen Dai, Xin Zheng, Fang Wang, Lei Guo
TL;DR
The paper tackles judicial sentencing prediction by balancing interpretability with predictive accuracy. It introduces the SMNN framework, a hybrid of a mechanism-based sentencing model and a neural compensator, and a two-stage learning strategy that first uses adaptive stochastic gradient (ASG) to initialize the mechanistic parameters and then applies Adam to refine all parameters. The authors prove global asymptotic convergence for the ASG stage without excitation data and demonstrate, on a real-world intentional injury dataset from China, that the SMNN with two-stage ASG-Adam yields superior predictive accuracy compared with saturated mechanism-only and neural baselines. The work highlights the value of integrating legal reasoning with data-driven adjustment and provides practical guidance for deployment in sentencing prediction tasks, with potential extensions to other crimes and further theoretical guarantees for the second-stage optimization. In particular, the use of saturation bounds, linear-in-parameters reformulations, and a bias-consistent loss in the second stage contribute to robust, interpretable, and accurate predictions.
Abstract
The investigation of legal judgment prediction (LJP), such as sentencing prediction, has attracted broad attention for its potential to promote judicial fairness, making the accuracy and reliability of its computation result an increasingly critical concern. In view of this, we present a new sentencing model that shares both legal logic interpretability and strong prediction capability by introducing a two-stage learning algorithm. Specifically, we first construct a hybrid model that synthesizes a mechanism model based on the main factors for sentencing with a neural network modeling possible uncertain features. We then propose a two-stage learning algorithm: First, an adaptive stochastic gradient (ASG) algorithm is used to get good estimates for the unknown parameters in the mechanistic component of the hybrid model. Then, the Adam optimizer tunes all parameters to enhance the predictive performance of the entire hybrid model. The asymptotic convergence of the ASG-based adaptive predictor is established without requiring any excitation data conditions, thereby providing a good initial parameter estimate for prediction. Based on this, the fast-converging Adam optimizer further refines the parameters to enhance overall prediction accuracy. Experiments on a real-world dataset of intentional injury cases in China show that our new hybrid model combined with our two-stage ASG-Adam algorithm, outperforms the existing related methods in sentencing prediction performance, including those based on neural networks and saturated mechanism models.