Software

Epidemiological modeling is used, under certain assumptions, to represent the spread of a disease within a population. Information generated by these models can then be applied to inform public health practices and mitigate risk. To provide useful and actionable preparedness information to administrators and policy makers, epidemiological models must be formulated to model highly localized environments such as office buildings, campuses, or long-term care facilities. In this paper, a highly configurable agent-based simulation (ABS) framework designed for localized environments is proposed. This ABS provides information about risk and the effects of both pharmacological and non-pharmacological interventions, as well as detailed control over the rapidly evolving epidemiological characteristics of COVID-19. Simulation results can inform decisions made by facility administrators and be used as inputs for a complementary decision support system. The application of our ABS to our research lab environment as a proof of concept demonstrates the configurability and insights achievable with this form of modeling, with future work focused on extensibility and integration with decision support.

Refer to the following publication for more detail: P. Ciunkiewicz, W. Brooke, M. Rogers, and S. Yanushkevich, “Agent-based epidemiological modeling of COVID-19 in localized environments,” Computers in Biology and Medicine, vol. 144, p. 105396, May 2022, doi: 10.1016/j.compbiomed.2022.105396.

GitHub repository: https://github.com/PCiunkiewicz/localized-epidemiological-abs

Dynamic discretization toolkit in Python based on the work by Fenton and Neil. This library provides classes for performing both static and dynamic discretization on data, as well as a handful of custom Bayesian networks defined using PyAgrum for testing. There is additional functionality for visualization of the resulting discretizations.

GitHub repository: https://github.com/PCiunkiewicz/dynamic-discretization

Contains the models, dataset, and software needed to run the models from the Relatable Clothing papers.

GitHub repository: https://github.com/th-truong/relatable_clothing

Set of projects associated with decision support systems and case studies for developing trustworthiness in artificial intelligence.

Implementation of the following papers:

• K. Lai, S. Yanushkevich, V. Shmerko, and M. Hou, Capturing Causality in Bias in Human Action Recognition, in Pattern Recognition Letters, vol. 147, July 2021, pp. 164-171.

• K. Lai, S. Yanushkevich, and V. Shmerko, Epidemiology Attack at the Aircraft Carrier Theodore Roosevelt: Bridging Gaps of the Emergency Management, Journal of Defense Modeling and Simulation, July 2021, pp. 1-16.

• K. Lai, S. N. Yanushkevich and V. P. Shmerko, Intelligent Stress Monitoring Assistant for First Responders, in IEEE Access, vol. 9, February 2021, pp. 25314-25329.

• K. Lai, H. Oliveira, M. Hou, S. Yanushkevich, and V. Shmerko, Assessing Risks of Biases in Cognitive Decision Support Systems, European Signal Processing Conference, January 2021, pp. 840-844

• K. Lai, H. Oliveira, M. Hou, S. Yanushkevich, and V. Shmerko, Risk, Trust, and Bias: Causal Regulators of Biometric-Enabled Decision Support, IEEE Access, vol. 8, August 2020, pp. 148779-148792.

• K. Lai, S. Yanushkevich, and V. Shmerko, Reliability of Decision Support in Cross-spectral Biometric-enabled Systems, IEEE International Conference on Systems, Man, and Cybernetics, October 2020, pp. 1-6.

• K. Lai and S. Yanushkevich, Decision Support for Video-based Detection of Flu Symptoms, IEEE International Conference on Systems, Man, and Cybernetics, October 2020, pp. 1-7.

GitHub repository: https://github.com/rasgally/projects

This work addresses tone mapping, a common approach to convert high dynamic range (HDR) images into low dynamic range (LDR) images. We approach this problem by using adaptive tone mapping. We propose to deploy a conditional generative adversarial networks to build an adversarial and adaptive tone mapping operator (adTMO) that converts HDR into LDR images. We use an objective quality metric called the Tone Mapped Image Quality Index (TMQI) to evaluate our adTMO. Trained with 256∗256 images, adTMO is able to generate 256∗256 and high-resolution 1024∗2048 LDR images. Given 1024∗2048 HDR images, TMQI of the generated LDR images reaches the value of 0.90±0.06, which outperforms all other contemporary tone mapping operators.

GitHub repository: https://github.com/caoxingdong/adTMO

This work addresses two image-to-image conversions: conversion of a visible face image into a thermal face image (V2T), and one thermal face image into another one given a different target temperature (T2T). We propose to use conditional generative adversarial networks (cGAN) with cGAN loss, perceptual loss, and temperature loss to solve the conversion tasks. In our experiment, we used Carl and SpeakingFaces Databases. Frèchet Inception Distance (FID) is used to evaluate the generated images. As well, face recognition was applied to assess the performance of our models. For the V2T task, the FID of the generated thermal images reached a low value of 57.3. For the T2T task, we achieved a rank-1 face recognition rate of 91.0%.

GitHub repository: https://github.com/caoxingdong/Thermal2Thermal

DS-BN is a C++ executable that accepts input data related to probabilistic belief networks and Dempster-Shafer belief networks through the use of files. According to provided instructions from one of the files, performs computations and writes inferred probability distributions or Dempster-Shafer models  into an output file. Please, contact S. Eastwood for the manual and the .exe file (The executable only runs on Windows 7; to run the executable, delete the "_.txt" extension from the downloaded file).

DS-BN (version 2): download

DS-BN (version 2 manual): download

Example input files:
3_Variables_file.txt (Save as: 3_Variables_file.txt)
3_Factors_file.txt (Save as: 3_Factors_file.txt) (should contain a single character: 0)
3_DS_Factors_file.txt (Save as: 3_DS_Factors_file.txt)
3_Instructions_file.txt (Save as: 3_Instructions_file.txt)

Example output file:
3_Output_file.txt

GitHub repository: https://github.com/sceastwo/DS-BN-UofC

The "Causal Networks and Uncertainty Metrics" software package is a C++ executable that can process data stored in the format of decision diagrams. Various structures related to graphical uncertainty models, such as the large arrays that denote conditional probability tables and distributions, are stored in the format of decision diagrams. Decision diagrams carry the advantage of exploiting the symmetry present during context specific variable independence. Commands are present that can carry out uncertainty inference in the following metrics: point probabilities; triangular fuzzy probabilities; probability intervals; and Dempster-Shafer models. 

Github repository: https://github.com/sceastwo/Causal_Networks_and_Uncertainty_Metrics

SigGet - a Windows program for capturing signature from digitizing tablet (the tablet such as Wacom with 1024 levels of pressure must be installed on you PC). Coordinates, time, and pressure are recorded and saved to a Matlab MAT-file (C++) . Please, contact S.Yanushkevich for .exe file.
Function for approximating signatures with cubic splines (Matlab) 
Program for applying free-form deformation to a signature (with graphical user interface) (Matlab) 
Function for testing deformation of signatures using deforming polynomials (Matlab) 

A set of Matlab functions for  facial temperature estimation, based on IR imaging; was created for Matlab R2008. 

Package for generating fingerprints based on Gabor filter and Orientation map alternations (Matlab). The package was developed in 2005 for Matlab R2005, was not updated for the later versions of Matlab