Round 7 Holdout DatasetThis is the holdout data used to construct and evaluate trojan detection software solutions. This data, generated at NIST, consists of natural language processing (NLP) AIs trained to perform named entity recognition (NER) on English text. 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 8 Train DatasetThis is the training data used to construct and evaluate trojan detection software solutions. This data, generated at NIST, consists of natural language processing (NLP) AIs trained to perform extractive question answering (QA on English text. 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.

The library multicomplex is an implementation of multicomplex algebra in C++ to allow for higher-order derivatives of numerical functions. Many (though not all) mathematical functions are implemented, allowing for calculation of derivatives (straight and mixed) to approximately numerical precision, which is difficult or impossible to achieve in conventional double precision

Python module "optbayesexpt" uses optimal Bayesian experimental design methods to control measurement settings in order to efficiently determine model parameters. Given a parametric model - analogous to a fitting function - Bayesian inference uses each measurement "data point" to refine model parameters. Using this information, the software suggests measurement settings that are likely to efficiently reduce uncertainties. A TCP socket interface allows the software to be used from experimental control software written in other programming languages.

SEDCORR is an open-source Python module designed to correct for the systematic energy deficits in atom probe mass spectra of electrically insulating samples. The assumption of the algorithm is that the mass spectrum for a dataset is conserved throughout the dataset and that any changes to the peak positions arise from an unknown slowly-fluctuating accelerating voltage. For computational speed, the unknown accelerating voltage is determined using a template matching FFT-based cross correlation method.

Round 1 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 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.

SCATMECH is a library of object-oriented C++ computer codes originally developed for disseminating models for polarized light scattering from surfaces and aerosols and for diffraction from gratings. The pySCATMECH package has been developed as an interface to the SCATMECH library, simplifying use of the codes and allowing for more rapid development of software for these applications.

Python module 'optbayesexpt' uses optimal Bayesian experimental design methods to control measurement settings in order to efficiently determine model parameters. Given an parametric model - analogous to a fitting function - Bayesian inference uses each measurement 'data point' to refine model parameters. Using this information, the software suggests measurement settings that are likely to efficiently reduce uncertainties. A TCP socket interface allows the software to be used from experimental control software written in other programming languages.

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.