This report describes the test equipment, process, analysis, and data file formats for the Numisheet 2020 tensile and tension/compression testing of the four materials associated with the Numisheet 2020 conference benchmarks. The four materials include two steel alloys, DP980 for Benchmark 1 and DP1180 for Benchmark 2, and two 6000 series aluminum alloys, AA6xxx-T4 for Benchmark 1 and AA6xxx-T81 for Benchmark 2, that will be referred to here as BM1-DP980, BM2-DP1180, BM1-6xxx-T4, and BM2-6xxx-T81, respectively.

Round 1 Test DatasetThe data being generated and disseminated is the test data used to evaluate trojan detection software solutions. This data, generated at NIST, consists of human level AIs trained to perform a variety of tasks (image classification, natural language processing, etc.). A known percentage of these trained AI models have been poisoned with a known trigger which induces incorrect behavior. This data will be used to develop software solutions for detecting which trained AI models have been poisoned via embedded triggers.

Round1 Holdout DatasetThe data being generated and disseminated is the holdout data used to evaluate trojan detection software solutions. This data, generated at NIST, consists of human level AIs trained to perform a variety of tasks (image classification, natural language processing, etc.). A known percentage of these trained AI models have been poisoned with a known trigger which induces incorrect behavior. This data will be used to develop software solutions for detecting which trained AI models have been poisoned via embedded triggers.

Round 2 Training DatasetThe data being generated and disseminated is the training data used to construct trojan detection software solutions. This data, generated at NIST, consists of human level AIs trained to perform image classification. A known percentage of these trained AI models have been poisoned with a known trigger which induces incorrect behavior. This data will be used to develop software solutions for detecting which trained AI models have been poisoned via embedded triggers.

This Standard Reference Material (SRM) is in the form of chips sized between 0.50 and 1.18 mm sieve openings (35 and 16 mesh). It is intended for use primarily in chemical methods of analysis. Similar material for use in spectrometric methods of analysis is available as SRM 1297. This data is public in the Certificate of Analysis for this material.

Between 1922 and 1940, the National Bureau of Standards (NBS) conducted a long-term investigation into the corrosion of bare steel and iron, pipes buried underground at 47 sites representing different soil types in the United States. Following the passage of the 2004 Pipeline Safety Improvement Act, the Department of Transportation's Office of Pipeline Safety requested that NIST review and reanalyze the data from the NBS study using modern statistical analysis tools. For this analysis, NIST compiled an updated database from the data in the publications by K. H. Logan (i.e.

The National Institute of Standards and Technology, as part of its report on the collapse of the World Trade Center, characterized many important steels recovered from the buildings to provide stress-strain models to analyze the impact, fires, and resulting collapse. Those tests represent a large additional data set that can be used for modeling the response of steel structures to fire.

Round 3 Training DatasetThe data being generated and disseminated is the training data used to construct trojan detection software solutions. This data, generated at NIST, consists of human level AIs trained to perform image classification. A known percentage of these trained AI models have been poisoned with a known trigger which induces incorrect behavior. This data will be used to develop software solutions for detecting which trained AI models have been poisoned via embedded triggers.

Round 2 Test DatasetThe data being generated and disseminated is the test data used to evaluate trojan detection software solutions. This data, generated at NIST, consists of human level AIs trained to perform a variety of tasks (image classification, natural language processing, etc.). A known percentage of these trained AI models have been poisoned with a known trigger which induces incorrect behavior. This data will be used to develop software solutions for detecting which trained AI models have been poisoned via embedded triggers.

Round 2 Holdout DatasetThe data being generated and disseminated is the holdout data used to evaluate trojan detection software solutions. This data, generated at NIST, consists of human level AIs trained to perform a variety of tasks (image classification, natural language processing, etc.). A known percentage of these trained AI models have been poisoned with a known trigger which induces incorrect behavior. This data will be used to develop software solutions for detecting which trained AI models have been poisoned via embedded triggers.