# Towards a Better Microcredit Decision

arXiv: https://arxiv.org/abs/2209.07574 · HTML: https://mlsanity.com/paper/2209.07574/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 9 of its 39 references that are in the map.
- First posted on arXiv 23 Aug 2022.

## Builds on (most cited first, 9 of 9)

- [XGBoost: A Scalable Tree Boosting System](https://mlsanity.com/paper/1603.02754/) (2016)
- [Cross-stitch Networks for Multi-task Learning](https://mlsanity.com/paper/1604.03539/) (2016)
- [Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Rate](https://mlsanity.com/paper/1804.07931/) (2018)
- [Attentional Factorization Machines: Learning the Weight of Feature Interactions via Attention Networks](https://mlsanity.com/paper/1708.04617/) (2017)
- [Trace Norm Regularised Deep Multi-Task Learning](https://mlsanity.com/paper/1606.04038/) (2016)
- [Modeling the Sequential Dependence among Audience Multi-step Conversions with Multi-task Learning in Targeted Display Advertising](https://mlsanity.com/paper/2105.08489/) (2021)
- [Deep Generative Models for Reject Inference in Credit Scoring](https://mlsanity.com/paper/1904.11376/) (2019)
- [Shallow Self-Learning for Reject Inference in Credit Scoring](https://mlsanity.com/paper/1909.06108/) (2019)
- [Deep Bayesian Multi-Target Learning for Recommender Systems](https://mlsanity.com/paper/1902.09154/) (2019)

## Built on by (most cited first, 2 of 2)

- [Best Practices for Responsible Machine Learning in Credit Scoring](https://mlsanity.com/paper/2409.20536/) (2024)
- [Balancing Performance and Reject Inclusion: A Novel Confident Inlier Extrapolation Framework for Credit Scoring](https://mlsanity.com/paper/2510.12967/) (2025)

## Abstract (arXiv)

Reject inference comprises techniques to infer the possible repayment behavior of rejected cases. In this paper, we model credit in a brand new view by capturing the sequential pattern of interactions among multiple stages of loan business to make better use of the underlying causal relationship. Specifically, we first define 3 stages with sequential dependence throughout the loan process including credit granting(AR), withdrawal application(WS) and repayment commitment(GB) and integrate them into a multi-task architecture. Inside stages, an intra-stage multi-task classification is built to meet different business goals. Then we design an Information Corridor to express sequential dependence, leveraging the interaction information between customer and platform from former stages via a hierarchical attention module controlling the content and size of the information channel. In addition, semi-supervised loss is introduced to deal with the unobserved instances. The proposed multi-stage interaction sequence(MSIS) method is simple yet effective and experimental results on a real data set from a top loan platform in China show the ability to remedy the population bias and improve model generalization ability.