follow us on facebook follow us on twitter email us

Materials Science and Engineering Data Challenge

About the Challenge
We are seeking novel uses of accessible digital data to advance Materials Science and Engineering knowledge to accelerate the transition to industrial applications

Posted By: Air Force Research Lab
Category: Scientific/Engineering
Partnership With: National Institute of Standards and Technology
National Science Foundation
Submission Dates: 12 a.m. ET, Jul 01, 2015 - 11:59 p.m. ET, Mar 31, 2016 Judging Dates: Apr 01, 2016 - May 11, 2016 Winners Announced: May 11, 2016

People have successfully applied advanced analysis techniques to ‘big data’ and have solved significant challenges in several areas of society including business, biology, and astronomy.  Using these techniques has provided people with new and valuable information from the data that was not readily available by other means.  However, materials science and engineering data has not yet been exploited to its full potential because of its complexity.  This complexity also stands to provide rich insights if the mysteries the data hold can be unraveled.

To advance the National goals of the Materials Genome Initiative (MGI), we are soliciting innovative approaches to solve materials science and engineering problems primarily through the analysis of publicly accessible digital data. Areas of particular interest include discovery of new materials to meet an application need, or development of a new model describing processing-structure-property relationships in either a structural (load bearing), functional (electrical, optical or magnetic), or multifunctional material.

Emphasis will be on use of existing and accessible data sources, novelty of the approach, and validation of results.

Winners will be invited to present their work at a special session (reasonable expenses paid) during Materials Science & Technology 2016 in Salt Lake City, Utah.

This challenge requires only a written research report in a format suitable for peer-reviewed scientific publication be submitted. The challenge award will be contingent upon evaluation of the proposal by the review committee.  The Air Force Research Laboratory intends to make multiple awards with the top award being $25K and no award less than $5K depending on creativity and completeness of the report.

To receive an award, the Solvers will not have to transfer their exclusive IP rights to the US Government.  Instead, they will grant to the US Government non-exclusive license to practice their solutions. Other arrangements may be considered, if a license cannot be granted.

The US Government is soliciting advanced approaches to analysis of digital materials data that would provide new knowledge for materials science and engineering to a level where they would enable:

  1. Discovery of new material structures and/or compositions that provide previously unattained and useful properties that enable new industrial applications or products.
  2. Development of new models to link at least two elements in materials processing-structure-property relationships where those relationships are currently poorly described by physics-based models.

Note: Solvers can be individuals or teams from academia, industry, or any other area. Solvers can submit reports for solutions to either or both of the needs described in 1 & 2 above, but they should submit a separate report for each if responding to both. Reports must detail the source of the materials data, the research approaches used, and methods used to validate the findings.

We are not seeking submissions for the following areas: Research in biology for pharmaceutical, medical, or other biological or biomedical purposes.

Background

Reuse or repurposing of digital scientific data has been applied in many scientific disciplines to solve important technical challenges for that discipline. Advances in computational power and data science techniques have enabled these innovations by providing new means to extract information from large datasets. Materials science and engineering progress is dependent on understanding, connecting, and applying extraordinarily complex information across large temporal and spatial scales. However, researchers often times encounter hard limits on the availability of accurate physical models to describe behavior. This lack of well-developed physics-based approaches across scales hampers advances in the discovery of new materials and an ability to link processing-structure-property relationships in such a way as to enable quantitative and predictive materials design tools. The goal of the Materials Genome Initiative is to halve the time it takes to discover, develop and deploy a new material and at a fraction of the cost compared to traditional approaches. Application of data science techniques to accessible digital materials data to develop new knowledge is expected to be part of the solution to reaching this goal.

The goal of this project is to leverage advanced computational and data science techniques to solve challenges associated with discovery, development, and production of new materials. The US Government is soliciting innovative, detailed approaches that can solve barrier issues in materials science and engineering research and offer new avenues for providing rapid and insightful information in materials research and development.

Previously unpublished materials data may be submitted here (subject to NIST approval):

https://materialsdata.nist.gov/

More information on the Materials Genome Initiative can be found at:

http://www.mgi.gov/

OSTP article on the Materials Genome Initiative can be found here:

OSTP article on MGI

 

 

 

 

Judges
Evaluation of the solutions will be conducted by a panel of scientific experts in materials science and engineering.
Judging Criteria

The materials data used in the challenge is accessible, discoverable, and has sufficient provenance

• Data used in this challenge must be publicly accessible on the internet by 1 Sep 2015, and documentation of this accessibility should be provided with submission. If desired, documentation of data availability can be established by sending a short description of the data and the accessible link to challenge@wbi-icc.com no later than 1 Sep 2015.
• Data must be discoverable and cited through a means such as a resolvable Uniform Resource Identifier (URI) or, ideally, a Persistent Identifier (PID).
• Data sources used must have sufficient supporting information (metadata) to enable reuse by researchers outside the original research effort.
• While the data used is not required to be free, or licensed for unrestricted use, cost and license encumbrances will be considered during the judging to ensure that the data is not effectively inaccessible to large swathes of the research community.
• Greater weight will be given to solutions that reuse data generated by others than the solvers themselves.

If hosted on a separate site, the the link must be part of the submission.
This criteria must be met for a valid submission.

Novelty and significance of the approach

• The approach is clearly defined through a written report suitable for peer-reviewed scientific publication and makes a significant contribution to the goals of this challenge.
• The solver has employed data analysis techniques that significantly accelerate the attainment of new insights and information as compared with existing experimental or physics-based modeling means.
• If the approach identifies a new material composition, material structure, or family of compositions, the solution will be evaluated against the degree to which the predicted and/or demonstrated range of properties provide significant improvement over currently known materials for a specific or variety of industrial applications. An analysis of the ability for or actual demonstration of practical synthesis of the material composition should be addressed in the report.
• If the approach addresses modeling the linkage between at least two elements of material processing, structure, and/or property, a quantitative description of material structure (e.g. microstructure) must be included in the model. The solution will be evaluated against the degree to which the new physical insight improves gaps in our ability to accurately model material behavior across length and/or temporal scales.

Validity of solution

• Scientifically sound steps have been taken and described to demonstrate that the approach has produced a valid solution, within a reasonable confidence level.

Uniqueness of approach or model

• The approach is primarily based on using data analytical techniques (e.g. machine learning, Bayesian Neural Networks, etc.) with existing accessible data to derive new information in materials science and engineering.
• The results of the approach must not have been previously publicly presented nor published

This is a disqualifying criteria.

How to Enter

Fill out the submission form for personal contact information.

Submit the solution as an attached file in the attached solver submission format without any personally identifying information (all entries will be anonymous to the judging team)

Submission form

Previously unpublished data can be published in the NIST Materials Data Repository:

NIST Materials Data Repository

 

 

Prizes
Materals Genome Iniatiative Prize $50,000.00 The awards will be paid to the best submission(s) as solely determined by the review committee. The US Government intends to make multiple awards with the top award being $25,000 and no award being smaller than $5,000. • In addition to the monetary awards, winning Solvers will be invited to participate (present their research and engage the audience in a question and answer session) in a conference session at Materials Science & Technology 2016, to be held October 24-27, 2016, Salt Lake City, Utah. Travel expenses associated with attending and presenting at the conference will be covered up to a maximum of $3,000 per submission.

29 Discussions for "Materials Science and Engineering Data Challenge"

    • Reply
      rlee
      The Winners have been announced. The winners were the following teams led by Winner Award Winning submission title Joshua Gomberg $25,000 Structure-based Energy Models from Simulated Al Grain Boundary Datasets Wenhao Sun $5,000 The Thermodynamic Scale of Inorganic Crystalline Metastability Ayman Salem $5,000 Cloud-Based Data-Driven Modeling for Fatigue Life Prediction in Ti-6Al-4V for Turbine Blade Applications Theron Rodgers $5,000 Extraction of Process-Structure Linkages from Simulated Additive Manufacturing Microstructures Using a Data Science Approach Bruno Abreu Calfa $5,000 Optimal Design of Atomic Crystalline Solids using Kernel Regression Property Prediction Models   Anton Oliynyk $5,000 Deceptively Simple and Endlessly Complicated: Machine Learning Prediction and Experimental Confirmation of Novel Heusler Compounds

  • Show Replies [+]
    dbrough3
    The website indicated that the submission deadline is today. I just tried to submit a solution but was unable to and found a message stating the submission period had ended. Can I not submit a solution today?

    • Reply
      Bruno Abreu Calfa
      I am having the same problem. In the submission page, there was a part saying: "Submissions to this Challenge must be received by 11:59 PM (US Eastern Time) on 31-Mar-2016. Late submissions will not be considered." So I am confused why it is closed now (March 31st, 10:55 AM EST).

  • Show Replies [+]
    Anton Oliynyk
    Hello Dear Organizers, I am wondering how to submit our work for both of the needs. 1. Discovery of new material structures and/or compositions that provide previously unattained and useful properties that enable new industrial applications or products. 2. Development of new models to link at least two elements in materials processing-structure-property relationships where those relationships are currently poorly described by physics-based models. Do we just need to submit it two times and each time indicate in the description where it should go? Thanks for your help, Regards,

  • rfed
    Hello, we were wondering if we we (or anyone else) could be granted a 1-2 week extension on the submission deadline? It would be quite helpful, if possible..

    • Reply
      Robert Lee
      Not sure what this discussion line is about. Can you clarify what module you are referring to?

  • rlee
    A clarification on the previous discussion. the Eligibility rules are that only AFRL, NIST, and NSF Government employees are not eligible for the prize since they are the sponsors of the challenge. Therefore employees of any other government agency are eligible.

  • rlee
    Question: I work for EPA, but this challenge does not cover what I do for EPA. I have not heard anything from your group apart from the automatic message that includes ”someone will respond once the feedback has been reviewed.” EPA is still not specifically mentioned in the eligibility section of the challenge announcement, so I was hoping that only the employees from the agencies funding this challenge would be denied potential award money. Answer: This is fine you are eligible to enter.

  • rlee
    Question: Can we cite a paper that is under review and that describes parts of the methodology? I can't guarantee that the paper will be publicly available by the submission deadline, and the idea of repeating much of the methodology in the submission doesn't seem appealing. When submitting a solution, can I upload a draft of the paper under review or create a link to it in case it hasn't been published yet? Any comments on how to go about this? Answer: It's fine to either provide a link to the referenced paper, or upload it along with the paper submitted for the challenge (as an appendix if needed).

  • Bruno Abreu Calfa
    Can we cite a paper that is under review and that describes parts of the methodology? I can't guarantee that the paper will be publicly available by the submission deadline, and the idea of repeating much of the methodology in the submission doesn't seem appealing. When submitting a solution, can I upload a draft of the paper under review or create a link to it in case it hasn't been published yet? Any comments on how to go about this?

  • rlee
    We were notified that there was a submission deadline error in the write-up. this has been corrected and should now consistently state that the deadline for submissions is 31 Mar 2016. Thank you to Ashley for bringing this to our attention.

  • rlee
    Question: Can the solution be submitted to a journal after the results come out? I am wondering if by submitting a solution to this challenge will eliminate the possibility to submit a full article to a journal later on Answer: the submission of a solution to this challenge is a private communication between the solver and the evaluation team. This should not affect your ability to submit your solution to a peer reviewed Journal. In Fact after you submit a solution to this challenge, we would prefer that you publish the work in a peer-reviewed publication. If you are a winner we would like to create a short synopsis of your work and advertise your success.

    • Reply
      rlee
      Answer: the submission of a solution to this challenge is a private communication between the solver and the evaluation team. This should not affect your ability to submit your solution to a peer reviewed Journal. In Fact after you submit a solution to this challenge, we would prefer that you publish the work in a peer-reviewed publication. If you are a winner we would like to create a short synopsis of your work and advertise your success.

  • If you are interested in the goals of the Materials Genome Initiative, you may want to attend the TMS Introduction to Materials Informatics with Open Source Tools Workshop. Organized by The Minerals, Metals & Materials Society (TMS), the workshop will be held on the afternoon of Sunday, October 4, in conjunction with Materials Science & Technology 2015 (MS&T15). The workshop will allow participants to significantly augment their existing workflows involving microscale and continuum simulations or experimental characterization to enable more efficient processing of large ensembles of datasets. For more details, visit: http://ow.ly/Rc65h This workshop is sponsored by the TMS Materials Processing & Manufacturing Division and TMS ICME Committee, with support offered by AFOSR-MURI, NIST, and Georgia Tech’s IMAT.

  • rlee
    Question: Our manuscript is in the final stages of preparation for submission to a journal, but it has not been submitted yet. I anticipate that it will be published prior to the April 1st, 2016 judging date. Upon publication, we will want to present our work at relevant materials conferences. I am wondering how the "not previously published" rule applies to our situation. Answer: The review committee decided that previously published in this context means that the work could not be published prior to the launch of the challenge. The scenario outlined in the question is perfectly fine for submission to the challenge

  • rlee
    I am Robert Lee at the Wright Brothers Institute. we are acting as the neutral facilitators for this Air Force Challenge. Any future questions can be directed to me and I will coordinate with the judging team and post the answers on this discussion page. Please submit questions either on this discussion forum or directly to me at: bob.lee@wbi-icc.com

  • Show Replies [+]
    rlee
    Question: We are potentially interested in participating in the above named challenge in the ”development of new models to link at least two elements in materials processing-structure-property relationships”. We were wondering about the requirement to have our data publicly accessible by 1 Sept 2015. If the data we are using is solely found in published peer-reviewed articles that are already on the Internet (e.g., ACS Publications, APS Journals), is that sufficient for the purposes of this challenge? If not, should we host our curated data on Citrination or related sites? Answer: Since the data is publically available you just need to include a link to it in your submission, However hosting the curated data would help the judges evaluate your submission easier.

    • Reply
      rlee
      A further explanation of this from the sponsor is "As long as the sources of the data used are properly cited in the report, and have a publication date prior to 1 September 2015, then use of data extracted from peer-reviewed journals is permissible. Inclusion of a supplementary file containing the extracted data used in the study is recommended." I do concur that if the supplementary file was added it would be easier on the judges

  • rlee
    We had a question for all solvers: My group has questions regarding the Materials Science and Engineering Data Challenge. The posted challenge does not speak to citizenship requirements. Our group does have a member that is not a US citizen. Does this challenge require US citizenship status or is it open to all in academia? The Answer is the challenge is open to everyone.

  • Read here how Elsevier Materials Today is supporting the challenge in different way: http://www.materialstoday.com/materials-chemistry/news/materials-science-and-engineering-data-challenge/ including by offering computation time on the HPCC cluster: http://www.materialstoday.com/materials-chemistry/news/call-for-high-performance-computing-cluster-access/ Bat

  • Challenge participants: I thought I would link you to a two-part blog post I wrote about getting started in machine learning for materials scientists. See below--good luck! Part I: http://www.citrine.io/blog/2015/3/3/machine-learning-mat-sci-1 Part II: http://www.citrine.io/blog/2015/3/16/machine-learning-for-the-materials-scientist-feature-engineering-model-building

Add to the Discussion

Solutions View as List

1 Star2 Stars3 Stars4 Stars5 Stars (1 votes, average: 1.00 out of 5)
Loading...
Presents Creativity in Engineering-Science & Helping to ||Innovate||Creativity ||Design||Rese



Rules

1 Sep 2015        All data sources used must be published in accessible data repositories. Unpublished data can be submitted to the NIST materials repository at

NIST Materials Data Repository

31 Mar 2016     Final deadline for submitting a research report for consideration.

11 May 2016     Winners are announced

16 Oct 2016      Invited winners present their work at Materials Science & Technology 2016 in Salt Lake City, UT.

Project Criteria:

This challenge requires only a written research report in the format of a peer-reviewed scientific publication to be submitted. The submitted report should include the following:

    • A clear and concise description of the technical approach (methods) used to solve one of the two Challenges listed in the challenge description.
        • The approach should primarily be based on data analytic techniques (e.g. machine learning, Bayesian Neural Networks, etc.) using existing accessible data to derive new information in materials science and engineering.
    • A discussion of how this approach and solution compares against the state of the art.
      • If the approach identifies a new material composition, material structure, or family of compositions, the solution will be evaluated against the degree to which the predicted and/or demonstrated range of properties provide significant improvement over currently known materials for a specific or variety of industrial applications. An analysis of the ability for or actual demonstration of practical synthesis of the material composition should be addressed in the report.
      • If the approach addresses modeling the linkage between at least two elements of material processing, structure, and/or property, a quantitative description of material structure (e.g. microstructure) must be included in the model. The solution will be evaluated against the degree to which the new physical insight solves longstanding gaps in our ability to accurately model material behavior across length and/or temporal scales.
      • For either Challenge, the report should address steps taken to validate the solution. The solution will be evaluated against the degree to which commonly accepted means of validation were used (e.g. complementary experimental testing, use of training versus validation data, etc.).
      • The new techniques and/or knowledge generated in this work will be evaluated for their likelihood to significantly advance the state-of-the-art in materials science and engineering or provide a path forward for a new industrial product.
    • Citation(s) to the accessible data used in solving the Challenge and other relevant literature.
    •  All materials data used in the challenge must meet the following criteria:
      • Accessibility: data must be publically available via the internet. Solvers wishing to use their own materials data for analysis in this challenge must post the data in a data repository that is accessible.
      • Provenance: data sources used must have sufficient supporting information (metadata) to enable reuse by researchers outside the original research effort.
      • Discoverability: data must be discoverable and citable through a means such as a resolvable Uniform Resource Identifier (URI) or, ideally, a Persistent Identifier (PID).
  • The results of the approach must not have been previously publically presented nor published

The awards will be paid to the best submission(s) as solely determined by the review committee. The total payout will be $50,000, with at least one award being no smaller than $25,000 and no award being smaller than $5,000.

In addition to the monetary awards, winning Solvers will be invited to participate (present their research and engage the audience in a question and answer session) in a conference session at Materials Science & Technology 2016, to be held October 24-27, 2016, Salt Lake City, Utah. Travel expenses associated with attending and presenting at the conference will be covered by the Air Force Research Laboratory (AFRL) up to a maximum of $3,000 per submission.

All submitted ideas may be made publicly available (at the AFRL’s discretion) in their entirety, to foster open discussion and evaluation of the content generated by the Challenge.

  • To receive an award, the Solvers will not have to transfer their exclusive IP rights to the US Government , instead, they will grant to the US Government non-exclusive license to practice their solutions. Other arrangements might be considered, if a license cannot be granted.
  •  Submissions to this Challenge must be received by 11:59 PM (US Eastern Time) on 31-Mar-2016. Late submissions will not be considered.
  • After the Challenge deadline, the Seeker will complete the review process and make a decision with regard to the Winning Solution(s). All Solvers that submitted a proposal will be notified on the status of their submissions; however, no detailed evaluation of individual submissions will be provided.

Eligibility: Employees of the US Government  organizations (U.S. Air Force Research Laboratory, National Institute of Standards and Technology, and the National Science Foundation) may participate in the Challenge and receive Honorable Mention in the MS&T 2016 conference session but are ineligible to receive award money or travel funds to MS&T 2016.

Submit Solution
Challenge Followers
Public Profile: 0
Private Profile: 30