Automated Data Engineering Boosts FDM Warpage Detection Accuracy.

Saleh Valizadeh Sotubadi, Nazanin Mahjourian, Vinh Nguyen· July 22, 2026 View original

Summary

This study introduces an Automated Data Processing (ADP) framework, inspired by reinforcement learning, that optimizes machine learning model-feature combinations for predicting warpage in Fused Deposition Modeling (FDM). By using SHAP XAI for feature selection and a policy updating mechanism, the framework significantly improves predictive accuracy and stability.

Researchers have developed an Automated Data Processing (ADP) framework aimed at enhancing machine learning model performance for predictive tasks, specifically demonstrated in warpage detection for Fused Deposition Modeling (FDM). The core of this methodology is a reinforcement learning-inspired policy that iteratively evaluates and refines optimal combinations of machine learning models and feature sets. The framework trains multiple models on both full and SHAP-selected feature subsets across 217 datasets. It then assesses predictive accuracy and F1-scores, computes a scalar reward, and updates Q-values to guide future model and feature selection. This approach successfully converges towards optimal configurations, improving test-set AUC from 0.9248 to 0.9731 and increasing the mean reward value by over fifty percent compared to a baseline.

Why it matters

Manufacturing professionals can leverage this automated framework to improve the reliability and accuracy of quality control predictions, particularly in additive manufacturing, leading to reduced defects and optimized production processes.

How to implement this in your domain

  1. 1Explore integrating reinforcement learning-inspired frameworks for automated feature selection and model optimization in your ML pipelines.
  2. 2Apply SHAP XAI to identify and prioritize the most informative features for your predictive models, especially in manufacturing quality control.
  3. 3Develop a policy updating mechanism to continuously refine model-feature combinations based on performance metrics.
  4. 4Pilot this ADP framework on a specific manufacturing defect detection task to quantify its benefits in your environment.

Who benefits

ManufacturingAutomotiveAerospaceIndustrial Automation3D Printing

Key takeaways

  • An automated data processing framework significantly improves ML model accuracy and stability for warpage detection.
  • Reinforcement learning-inspired policy updating optimizes model-feature combinations.
  • SHAP XAI is effective for generating informative feature subsets, reducing dimensionality.
  • The framework achieved substantial improvements in AUC and mean reward compared to baselines.

Original post by Saleh Valizadeh Sotubadi, Nazanin Mahjourian, Vinh Nguyen

"arXiv:2607.18515v1 Announce Type: new Abstract: This study contributes toward development of an Automated Data Processing (ADP) framework designed to evaluate and reinforce optimal machine learning model-feature combinations for predictive tasks in fused deposition modeling (FDM)…"

View on X

Originally posted by Saleh Valizadeh Sotubadi, Nazanin Mahjourian, Vinh Nguyen on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses