Bayesian Dropout Approximation of Process Outcomes​

Project Lead: Rensselaer Polytechnic Institute

Member % Cost Share: 50%
CESMII % Cost Share: 50%
Duration: 18 Months

Problem Statement

The lack of robust, automatic predictive modeling is a gap that cuts across multiple industries. Traditional data-driven modeling techniques can fail to capture critical elements of processes and require significant domain expertise to exploit for process optimization and control.

Project Goal

  • Provide the CESMII Platform with a modeling engine with sophisticated predictive capabilities that can be invoked by Platform apps to model a variety of manufacturing processes.
  • Demonstrate the capabilities of that engine via a sensitivity analysis approach on a set of critical semiconductor manufacturing process data for electrical timing.

Leverage project outcomes for your manufacturing operations

Learn More

Technical Approach

  • This project will create a app, containerized and deployed on Azure, and primarily used by Platform apps as a modeling engine. It will interoperate with the SMIB to manage raw and processed data.

  • The app implements a stochastic neural network approach to capture complex relationships between data features in the form of output distributions. 

  • The app implements a basic suite of tools to postprocess distributions, to extract useful information and visualizations from model outputs. 

Key Tasks & Milestones

  • ​Design/Planning tasks:
    • API/Deployment definition (Q1)
    • Verification test suite definition and construction (Q1-Q2)

  • Implementation tasks:
    • Neural network code (Q2-Q4)
    • Code to package trained models (Q3)
    • First tier of model sampling tools (Q4)
    • First tier of post-processing/visualization tools (Q4)
    • Second tier of sampling and post-processing/visualization tools

  • Application tasks:
    • Verify all code against test suite (Q4)
    • Train network on IBM data set (Q4)
    • Execute analyses of IBM data sets (Q5)

  • Documentation and technology transfer (Q6)

Potential Impact

  • Applicable to very wide array of processes, with reusable, flexible models.
  • Capable of sophisticated predictive modeling
  • With low data requirements compared to traditional deep learning
  • That captures inherent variations of noisy processes
  • That captures “fat tail” behavior of non-Gaussian processes
  • That can use dark data sources to increase confidence in predictions of​ variety of quantities, including yields, resource usage, reliability, catalyst regeneration times
  • Capable of sophisticated process sensitivity analyses
  • Identify process gains and input interactions, leading to tighter process controls, both run-to-run and online.
  • Decrease waste, resource usage.


Addresses Target 2.1.4 from RFP: Modeling and Analytics, Hybrid modeling and application to thermal processes;  Addresses Target 2.1.5 from RFP: Smart Manufacturing Platform

Project Selection & Announcements