Sorry, you need to enable JavaScript to visit this website.

Machine Learning-Enhanced Energy-Product Development

Machine Learning-Enhanced Energy-Product Development
June 21-22, 2018 
Falls Church, VA

ARPA-E hosted a workshop entitled “Machine Learning-Enhanced Energy-Product Development” on June 21-22, 2018 in the Falls Church, VA area. 

The workshop convened leading experts in machine learning, artificial intelligence, topology optimization methodologies, inverse design methodologies, dimensionality reduction techniques, and general engineering design and optimization (including vendors of existing design tools and software), to identify innovations that have the potential to drastically improve the speed and quality of complex engineering design and thereby accelerate the development of next generation energy technologies whose discovery would otherwise be highly unlikely. Participants lent their expertise to help ARPA-E explore innovative technologies to determine relevant and compelling metrics that will define a successful research program. 

ARPA-E seeks to identify opportunities that can accelerate the pace at which energy technologies are developed and brought to market by employing emerging machine learning (ML) /artificial intelligence (AI) algorithms and toolkits. Together, both ML/AI and energy-technology experts will identify topic areas where investment in the application of ML/AI algorithms could yield either a dramatic enhancement in the performance of end-use technology (e.g. gas turbines, batteries, power converters), or a dramatic reduction in the cost (and risk) associated with the development of the technology. These above-mentioned topic areas may be technology-specific (e.g. materials, fluid dynamics) or development-stage specific (e.g. conceptual design, detailed design, due diligence). Potential approaches and targeted outcomes include, but are not limited to: 

Identify state-of-the-art in ML/AI tool kits, and the opportunities and challenges in leveraging them in engineering design
Identify opportunities/challenges and methodologies in dimensionality reduction techniques to reduce complexity in the design process
Identify opportunities/challenges and methodologies for enabling inverse design techniques, including topology optimization techniques.
Exploration and prioritization of a broad set of energy applications and problem statements
Identification of metrics and outcomes that would define a successful technology R&D effort.
Identification of methods for evaluating the “usability” of any design tools that could be developed through an R&D effort

Thursday, 6/21

12:30 - 12:45 PM Welcome and Introduction to ARPA-E
Conner Prochaska, Senior Advisor and Chief of Staff, ARPA-E

12:45 - 1:20 PM Machine Learning-Enhanced Energy-Product Development – Introduction and Workshop Goals
David Tew, ARPA-E Program Director

1:20 – 1:40 PM Text and Data Mining for Materials Synthesis
Guest Speaker: Prof. Elsa Olivetti, MIT

1:40 – 2:00 PM The Dark Reactions Project: A Case Study in Using Machine Learning to Accelerate Exploratory Materials Synthesis
Guest Speaker: Prof. Joshua Schrier, Haverford College

2:00 –  2:20 PM Learning to Computationally Design Engineered Systems
Guest Speaker: Prof. Mark Fuge, University of Maryland

2:20 - 2:40 PM Topology Optimization of Manufacturable Architected Materials and Components
Guest Speaker: Prof. Jamie Guest, John Hopkins University

2:40 – 3:00 PM Brief participant introductions

3:00 PM – 3:30 PM Break/Networking

3:30 – 5:00 PM Breakout Session 1: Opportunities, Challenges, and Potential Solutions

5:30 – 7:00 PM One-on-one meetings with Dr. David Tew, Program Director

7:00 PM Informal Networking – Organize on Your Own

Friday, 6/22

8:40 – 9:00 AM Day 1 Summary, Day 2 Objectives, and Q&A
David Tew, ARPA-E Program Director

9:00 - 9:20 AM What Google knows about machine learning languages and you may not (Spoiler: Only Swift and Julia Make the Cut)
Guest Speaker: Prof. Alan Edelman, MIT

9:20 – 9:50 AM Machine Learning: Issues and Opportunities
Guest Speaker: Dr. David Womble, ORNL

9:50 – 10:20 AM Don't Blink. AI in Industry and Future Opportunities
Guest Speaker: Dr. Michael Giering, UTRC

10:20 – 10:30 AM Break / Networking

10:30 AM – 12:00 PM Breakout Session 2: Potential Program Scope

12:30 – 2:30 PM One-on-one meetings with Dr. David Tew, Program Director