Machine Learning Underwriting for Fair Crowdsourced Investment

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing investment platforms lack diversified investment options, and automated underwriting systems face challenges with algorithmic bias, data quality, and regulatory compliance, leading to unfair treatment and increased costs.

Innovation Solution

A machine learning-based system that integrates fairness-aware techniques, bias audits, and human oversight to provide a comprehensive analysis of applicant behavior, offering a wide range of investment opportunities, including crowdsourced loans, while ensuring fairness and accuracy in underwriting decisions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If automated underwriting systems are implemented to process applications efficiently, then productivity and turnaround time are improved, but algorithmic bias and fairness issues arise

Engineering Contradiction:

Improveapplication processing efficiencyVSAvoidfairness and equity in underwriting decisions

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system implements continuous feedback loops where underwriting decisions are monitored for bias patterns, and the machine learning models are retrained with corrected data to eliminate detected biases, creating a self-improving system that maintains both efficiency and fairness

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

Human reviewers serve as intermediaries to audit and oversee automated underwriting decisions, providing a check on algorithmic bias while maintaining the overall efficiency of the automated system for high-volume processing

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If comprehensive data collection is performed to improve underwriting accuracy, then measurement precision is improved, but data quality issues and bias propagation occur

Engineering Contradiction:

Improveunderwriting decision accuracyVSAvoiddata quality and bias in training data

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The system performs preliminary data auditing and bias detection during the data collection and preprocessing phase, identifying and correcting biased or low-quality data before it enters the training pipeline, preventing bias propagation to downstream decisions

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system dynamically adjusts data selection parameters and weighting factors based on detected bias patterns, modifying which data points are emphasized or de-emphasized in training to counteract identified biases while maintaining predictive accuracy

Inventive Principle:

Principle #35Parameter changes

3Reliability

If fairness-aware machine learning techniques are implemented to address bias, then fairness and equity are improved, but system complexity and costs increase

Engineering Contradiction:

Improvefairness in underwriting decisionsVSAvoidsystem complexity and implementation costs

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The fairness assurance system is segmented into modular components (bias detection module, data auditing module, model retraining module) that can be independently implemented and scaled, allowing institutions to start with basic fairness checks and add more sophisticated techniques as needed

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250217892A1Systems and methods for automating crowdsourced investment processes using machine learning

Publication Date: 2025.07.03 PNC FINANCIAL SERVICES GROUP INC

AI summary

The present disclosure describes computer-implemented methods and systems for automating application processing with dynamic data collection and augmentation related to applicants' behavior. The method includes receiving a plurality of applications with corresponding information, then aggregating, storing, and preprocessing data related to the applicants' behavior. The method also includes a machine learning model including a training dataset, a feature selection module, a hyperparameter tuning module, and a prediction model. The method includes predicting the application outcome based on the application information and behavior data and generating an underwriting decision based on the prediction. The method further includes providing underwritten applications for display and receiving a selection of a subset of the underwritten applications.