Options

Projects
Co-Authors
Relations
Faculty

• C-Apps9 - FinTech4Society

  FinTech4Society is a team of academics, united by one common goal: to understand what AI can bring to the emerging field combining finance and technology known as Fintech. A main topic is the interplay between algorithmic decisions and the socio-economic environment, as illustrated by the following examples:

  1. Recent advances in natural language processing are used more and more in financial markets.
  2. In some markets, financial institutions may have access to sophisticated algorithms, while households do not have such possibilities.
  3. New educational technologies are gaining ground in industrial practices.
  4. When new algorithms are developed, privacy of individuals may not always be adequately protected.

  The following research questions are of particular importance:

  1. How do outcomes of algorithms affect financial markets and the economy?
  2. How does a machine think differently from a human and what are the consequences of this difference?
  3. How does bias in machine learning and the underlying data affect outcomes?
  4. How do economic agents value privacy and how does AI affect this valuation? 

  Thanks to the various backgrounds of the team members, these questions will be tackled using a range of different methodologies, from AI techniques to empirical methods to theoretical analyses.

• C-Apps10 - Machine Learning In Physics

  Both machine learning and deterministic algorithms are used in the Physical and Engineering Sciences. What are the advantages and shortcomings of each approach for applications in the natural and applied sciences? Is there potential for hybrid algorithms? Can we devise physics-informed machine learning algorithms, or machine-learning-enhanced deterministic modeling and inversion?

• C-Cult5 - AI^4 Sound Societies (AI4S2): Adaptive Sonic Environments for Well-Being

  How to generate an adaptive sonic environment that enhances wellbeing? 

  This problem--sometimes subsumed under the general notion of “smart” or “intelligent” environment-- has been addressed in several literatures, from acoustic ecology and music therapy, to health science, to mechanical and civil engineering.  But as AI has developed by leaps and bounds in recent years, its applications to customizing sonic environments have lagged.

  Our collaboratory’s focus is AI-driven systems for adaptive soundscapes: linking people, environments, and algorithms in communications loops, using feedback to customize an auditory environment enhancing inhabitants’ social wellbeing. Potential components of such systems include computers and algorithms, biosensors (tracking physiological data, or spatial position), environmental sensors (microphones, thermometers, barometers), big datasets (including historical profile data on system users), and sound generators (speakers or headphones).

   The roles of AI include recognizing and interpreting bio- and environmental signals (e.g. using supervised learning), and developing effective responses (e.g. using reinforcement learning), whether by selecting, tuning, and mixing pre-recorded sounds, or by generating new ones (e.g. using Generative Adversarial Networks). Biosignal interpretation includes AI-mediated sonification, enabling users to learn to control their own autonomic systems.  Such systems may draw upon, as well as generate, big datasets, which are important for supervised learning.

• C-Cult6 - The AI Arts & Culture Salon

  Our research group is centered around a monthly salon in which we read current AI Arts & Culture Research. The goal of this group is to foster a congenial and critical culture at the UA around the role of AI in the wider world.

  Find workshops organized in 2021 by this collaboratory, here: 

  • AI Arts & Cultures Salon Workshop Series
  • Becoming Your ‘Authentic’ Self: How New Media Constructs Youth’s Visual Transitions (AI Arts & Cultures Salon Workshop Series)
  • Distributing Blackness (AI Arts & Cultures Salon Workshop Series)
  • The Algorithm of Long Algorithmic Culture (AI Arts & Cultures Salon Workshop Series)
  • On Information Activism: A Queer History of Lesbian Media Technologies (AI Arts & Cultures Salon Workshop Series)
  • Why the Data Revolution Hasn’t Happened Yet (AI Arts & Cultures Salon Workshop Series)
  • Fact Signaling, and AI's Certainty Fetish (AI Arts & Cultures Salon Workshop Series)

  Find the conferences and talks organized in 2022 by this collaboratory, here: 

  • Dr. Jemma DeCristo (UC Davis) | February 7th - 3:00pm | "Blackness against the Surface of Black Art"
  • Dr. David Marriott | March 8th - 3:00pm | "Of Effacement"
  • Dr. Neferti X.M. Tadiar | March 28th - 3:00pm | "A Physics Lesson: Notes on a Cultural Genealogy of Human Mediatic Forms"
  • Dr. Luciana Parisi | April 5th - 2:00pm | Artificial "Epistemologies in Computational Thinking"
• C-Cult7 - Know Thyself as a Virtual Reality

  Know Thyself as a Virtual Reality is an interdisciplinary project that focuses on the ethics and aesthetics of the use of medical data and virtual reality. In order to explore this field, we have assembled a skilled and diverse team from across the University of Alberta as well as from other Canadian and international universities, bringing together expertise in medical imaging, health law, software development, biomedical engineering, virtual narratives, sound, the history of medicine, and artistic creation. This collaborative team explores, creates, disseminates and share ideas about cutting-edge technologies that produce bodies in virtual and augmented reality. These bodies offer striking opportunities for producing new medical knowledge, but also render bodies transparent, accessible and manipulable in potentially exploitative ways, particularly during an era of increasing surveillance. Our project addresses these and other issues, allowing scholars and students to experiment with the technologies, learn from each other, and engage a broader public. This unique fusion of medical diagnostics and fine arts will ultimately help us to better know ourselves and our bodies in the digital age. 

  As part this research project we engaging with AI and ethics in several ways. Firstly, the team is creating a VR project that allows users to manipulate scan datasets, some are open source downloaded from the Internet (and have been anonymized automatically), and others have been acquired specifically for the project. Into this project will be placed an AI avatar that will learn to ‘assist’ the users.  Secondly, we are organizing an Art and AI speaker series for Fall 2021/Winter 2022 that will showcase the work of artists are working with AI both practically and creatively. Finally, we are creating a suite of software tools and ethical guidelines to allow other researchers to create VR experiences with medical scan data and AI. These tools will be offered to artists to make VR artworks for a public exhibition in 2022. 

• C-Cult8 - AI4Buzz: Pulse Check for Canada’s Energy Conversations

  Canada is facing challenges in innovation, energy policy, and environmental discussions. Yet, the public's trust in their institutions - government, business, and media – is rapidly eroding as is social license to operate/govern and market access. For the first time, “a person like me” has become as trusted as an academic or technical expert and twice as trusted as a board director, government official or CEO. Effective risk communication is fundamental to addressing sustainability challenges such as supporting energy transition/transformation, developing climate change policies, and enhancing necessary social engagement and innovation. This requires an interdisciplinary, collaboratory approach to understand how energy systems are being discussed, how the associated risks are being assessed, how risk acceptability is evaluated, and other overarching principles for effective communication: determining key audiences, defining the purpose of a risk communiqué, using the most effective method of communicating (visual, verbal, statistics, combinations), and using the most credible messenger. By applying quantitative text and network analysis techniques, we are mapping diverse knowledge domains to understand how individuals create shared meaning, collaboration networks, and innovations.

  ____________

  NEWS

  Project Update A Toolkit for Analyzing Online Conversations for Solutions Based Policy Development

  We’re excited to share the amazing work on this project by announcing that the platform is now LIVE to the public!  

  Visit here: http://ai4buzz.ca

  About the tool

  Social media platforms are public venues where conversations about issues of public interest take place. These online conversations have real world impact. The project objective is to understand the vocabularies, speakers, interlinking of conversations, associated topics/themes, and make this all public to draw these narratives together and support solutions-based policy conversations.

  The tool can aggregate these indicators at the level of tweet authors to shed light on the activities and style of influencers of public opinion. Finally, it offers rich visualizations to enable users to gain insights on their datasets. 

  This is public research project. Everyone has equal access to these conversations. Everyone can suggest an examination of a new energy type, new visualizations, and new analysis. This research is by Canadians and for Canadians to support conversations about Canadian energy policy. By helping us make this project better, you will enhance the accessibility and relevancy of energy policy to all Canadians. So, please make suggestions: http://ai4buzz.ca/suggestions

  The project leveraged expertise from engineering, computer science and business faculties and was peer reviewed by the University of British Columbia, University of Ottawa, University of New Brunswick, Cornell University, and Stanford University. The project is enabled through the generous donation of computing infrastructure through AI-HUB, University of Alberta Signature Area AI4Society, and support from the Mitacs Accelerate Program.  Researchers included: Drs. Lianne Lefsrud, Eleni Stroulia, Denilson Barbosa, Joel Gehman and Candelario Gutierrez, Andrea Whittaker, Ruby de Jesus, and Katherine Patenio.

  How to use the tool - Tutorials

  #energy #canada #news #tools #toolkit #analysis #innovation #climatechange #climate #environment #technology #tech #energyefficiency #energytransition #socialmedia #conversations #data

  ____________

  ARTICLES

  Gutierrez, C. A. G., Whittaker, A., Patenio, K. M., Gehman, J., Lefsrud, L. M., Barbosa, D., & Stroulia, E. (2021, November). Analyzing and visualizing Twitter conversations. In Proceedings of the 31st Annual International Conference on Computer Science and Software Engineering (pp. 4-13).

• C-DHLTH4 - Computational Psychiatry

  We are a multidisciplinary group based in Edmonton, Alberta, Canada, including scientists and clinicians who work in the field of machine learning for mental health. 

• C-DHLTH5 - Machine Learning to Rapidly Detect Drug-Resistant Human Fungal Pathogens

  Healthcare professionals must quickly identify pathogens to initiate prompt antimicrobial therapy to patients and to contain their spread within the hospital and community at large. However, drug-resistant pathogens can be notoriously difficult and time consuming to identify using traditional diagnostic tests. We are developing machine learning models to detect drug resistance in fungal pathogens from patient samples. These machine learning models are anticipated to detect drug-resistant pathogens more rapidly and accurately than the currently available diagnostic methods. The ultimate goal is to use machine learning to more effectively treat patients with drug-resistant infections and to better control the spread of infectious diseases.

• C-Eth3 - AI Ethics and Privacy

  With the dramatic advances in AI technology there has been a commensurate increase in public concern around the uses of AI in decision-making that affects people. The Ethics and Privacy Collaboratory would coordinate research and educational initiatives specifically around privacy and ethics issues related to big data and artificial intelligence.

   

  Events we've organized:

  Ethics Salons

  • Ethics Salon #11 - Kisha Supernant
  • Ethics Salon #10 - Amy M Schmitter, Howard Nye, Alona Fyshe & Matthew Guzdial
  • Ethics Salon #9 - Faith Majekolagbe and Meera Nair
  • Ethics Salon #8 - William Wadsworth
  • Ethics Salon #7 - Randy Gobel | Howard Nye and teams
  • Ethics Salon #6 - LeonardoNascimento
  • Ethics Salon #5 - Luke Stark
  • Ethics Salon #4 - Jeanna Matthews
  • Ethics Salon #3 - Hero Laird
  • Ethics Salon #2 - Jacob Jaremko
  • Ethics Salon #1 - Introductions

  Ethics Dialogues

  • Dialogue on the Ethics of Contact Tracing - Ethics Dialogues #1
  • Dialogue on Surveillance and Carcerality - Ethics Dialogues #2
  • Dialogue on Whistleblower Protections Can organizations self-regulate? - Ethics Dialogues #3
  • Ethics in the Age of Smart Systems - Annual Symposium

  Our Calls

  • Ethics in the Age of Smart Systems: Special Issue of IRIE
  • Symposium Call For Proposals: Ethics in the Age of Smart Systems
• C-IoT2 - Smart Buildings

  The digitization of building information has the potential to transform the way professionals design, construct and operate facilities and the way occupants interact with them. In particular, building information models (BIMs) enable the digital representation of a building’s physical and functional characteristics; and building automation systems (BASs) enable advanced building controls that can respond to occupant activities. The objective of the Smart Buildings collaboratory is to engage in research towards improving the planning, design, construction, operation, and maintenance of buildings, to increase occupant comfort and decrease energy consumption and costs. 

  A subproject of this Collaboratory, entitled Sex, gender and user studies in CS - Data usage, is led by Dr. Matthew Guzdial and investigates "inclusive space design".

• C-IoT3 - Crowdsensing-based Assessment of Wildfire-related Risks and Hazards

  In recent years, Canada has been experiencing an increased frequency of devastating wildfires. Especially in Western Canada, wildfires continue to occur with increased burn intensity and duration and have started to threaten urban areas. Thus, there is an urgent need to develop new methods for monitoring and assessing wildfire-related risks and hazards across broad geographic areas. These data are needed as inputs to modeling and forecasting systems that aim to protect the natural and built environments as well as public health and safety. In this context, the overarching long-term objective of this initiative is to develop a novel technology for assessment of wildfire-related risks and hazards through crowdsensing, which is a useful tool for obtaining information about various phenomena from large areas in nearly real time. We will develop a unique crowdsensing-based technology by employing crowdsourced image data from in-vehicle cameras analyzed with AI, computer vision, and image analysis with deep learning. We will study various phenomena including i) the exposure of critical infrastructure to wildfire by assessing the proximity of hazardous fuels to infrastructures and ii) important seasonal indicators that mark the beginning and end of the spring fire season known for human-caused fires: the timing of spring snowmelt and leafing-out of trees. It is expected that the framework will serve as a valuable tool for urban planning and fire management and will aid in the development of fire management plans by validating and enhancing observational data currently available from fixed sources such as fire lookout towers or fire weather stations.

• C-IoT4 - Human-centered Remote Assistance and Tele Presence for Automated Driving Systems

  Navigation of autonomous vehicles in highly dynamic urban environments are extremely challenging due to perceptually degraded situations and vehicle instability incidents, impacting decision-making. The project will develop and test a situation-aware learning control strategy, robust to time delays, for remote take-over and assistance for automated driving in complex environments.

  Automated driving systems (ADS) are leveraging advances in connectivity and sensing technologies to drive innovation in major markets across the globe. However, navigation of autonomous vehicles in highly dynamic urban environments, with human presence, are extremely challenging due to perceptually degraded situations (e.g., dynamic objects that deteriorate visual odometry, or dark/bright scenes) and vehicle instability incidents, impacting ADS model’s predictive capacity for decision-making. Remote take-over/assistance by a human operator provides a resilient decision-making replacement for controlling the vehicle and safe pullover in such situations. Designed from an interdisciplinary perspective, the project will develop and test a situation-aware learning control strategy, robust to time delays, for remote take-over and assistance for automated driving in complex environments, through the following three integrated objectives:

  
  O1. Situation awareness enhancement through visual-inertial remote sensing
  O2. Development of a remote human-centered control framework for shared autonomy
  O3. Implementation and experimental studies

• C-M1 - ELITE Analytics

  The objective of this project is to develop a data-driven evaluation methodology for mentoring programs, specially tailored to under-represented populations, such as the ELITE program.

• CD-CapB12 - Data Science Certificate

  Data literacy, the ability to derive meaningful information from data, is increasingly being recognized as a vital skillset for everyone in today’s data-rich world. Data analysis has always been the cornerstone of empirical science; today, data analysis has become at the very least useful but in many cases critical for many aspects of scholarly work. Data pours continuously in from the growing number of new devices that we are all using, social media, weather stations, environmental sensors, government agencies, credit card purchases, and browser logs. Most artifacts of our cultural activities today, including television, books, visual art, movies, and music, are born digital. Our ability to make sense of all this data, to decide on its authenticity and provenance, and to extract evidence that can drive decisions is a key skill for future professionals in most fields.  

  Many academic institutions, recognizing this phenomenon, are establishing academic credentials and programs for data science, both at the graduate and undergraduate levels. In Western Canada, UCalgary is launching a graduate program - https://science.ucalgary.ca/data-science, and SFU has an undergraduate program - http://www.sfu.ca/students/calendar/2020/fall/programs/data-science/major/bachelor-of-science.html

  The UoA’s strength in Machine Learning and Artificial Intelligence is well established as evidenced by international rankings, the support of the Pan-Canadian Strategy for AI to Amii, and the Administration’s support through the establishment of the AI4Society signature area. This makes the development of a data-science program perfectly timed, even long overdue. The proposed interdisciplinary embedded certificate is motivated by the intuition that we must tap into this expertise to gift our students today with the capacity to extract insights and knowledge from the data they will undoubtedly deal with in their profession in the future. 

• CD-CapB13 - Digital Law Modules

  This project will develop, test and publish three digital law modules to support digital literacy in Alberta’s legal communities. Each module will include an information component and an interactive component: digestible information on a key digital law topic, and ways for law students in particular, and lawyers and other community leaders more broadly, to consider those topics and how they might impact legal issues and services in Alberta.

• CD-CapB14 - Computer Vision with Active Learning for Social Good

  Learning Objectives: What would the user learn completing the module

  • Being comfortable with accessing and creating analysis pipelines for image-based data.
  • Learning how to clean/prepare/balance datasets for training ML models
  • Getting familiar with different Python tools for data analysis and visualization
  • Getting familiar with PyTorch/Torch Vision
  • Applying classification models to medical image and/or satellite data.
  • Applying segmentation models to medical image and/or satellite data.
  • Loading pre-trained image models and performing transfer learning.
  • Understanding what interactive machine learning is, and why/how it is important in social good applications (i.e. counting/classifying biodiversity data, providing labels for unlabeled data, and asking a park ranger to assist when it is uncertain about a classification)
  • Familiarity with some of the state-of-the-art literature for active learning/computer vision for social good.
• CD-CapB15 - Introduction to Voyant Guide

  This module will be designed to introduce text analysis using Voyant (www.voyant-tools.org). The module can be used by a teacher to teach a group of students and it can also be used by an independent learner. 

  A teacher teaching text analysis to students can adapt the exercises to suit the objectives of the class. A short PowerPoint presentation will be provided in Lecture 1 to help the teacher introduce Voyant to students. A teacher experienced with Voyant can replace the introduction videos by in class demonstration of the procedures covered in the videos.

  • Walkthrough 1: What is an electronic text and how can we study it?
  • Walkthrough 2: Studying a Text: The Arc of a Project
  • Walkthrough 3: Managing Corpora with Spreadsheets and XML
  • Walkthrough 4: Next Steps: Using Spyral Notebooks
• CD-CapB16 - Real-Time Tweets Processing & Analysis

  Objective 

  This tutorial is intended to give an introduction to the data stream context through a hands-on experience. We will work on the deployment of a modular and flexible real-time data processing pipeline that is going to be capable of materializing insights from tweets and generate visualizations for analysis.

  
  The tutorial will have 4 modules and a bonus analysis, as follows:

  Module 1: Twitter Developer Account Creation Process & Tweepy Client

  Module 2: Processing Tweets with Kafka 

  Module 3: Storing Processed Tweets in PostgreSQL 

  Module 4: Visualizing & Analysing Metrics with Superset Module 5: Development Considerations for a System in Production

  Bonus: Sentiment Analysis with NLTK Library

• CD-CapB17 - Development of a Course on “Artificial Intelligence for Biomedical Engineers” The main objective of the proposed initiative is to develop a course on “Artificial Intelligence for Biomedical Engineers” under a proposed accelerated Master of Engineering (M.Eng.) Program in Biomedical Engineering (BME) currently under development by the Department of Biomedical Engineering.
• CD-Eth11 - Three AI Ethics Case studies

  Case studies provide a concrete way to discuss ethical issues and concepts. There are three cases in this set of AI ethics tools that present local Alberta, real-world scenarios.

  1. On Demand Transit: This case explores how shifts in technology can have impacts on privacy in the context of the delivery of public services, such as public transit.
  2. Biometric Banking: This case explores the use of biometrics, a special class of data, in relation to providing banking services for underserved populations.
  3. Hey Mylo: Companion Robots: This case explores roboethics and issues related to human-robot relationships in the context of dementia care.                                                       

  The case studies can be used independently by students to reflect on the questions in each case with written responses. However, the cases are best applied in small group discussions. The exercises, in particular, are geared towards small groups of 4-6 students.

• CapB-P41 - Exploring global issues through big data: collaborating with teachers to strengthen data literacy in high school students through teaching mathematics for sustainable development This proposed project confronts mathematics teaching and learning through issues related to sustainability, data literacy and big data. For instance, There is no shortage of global issues requiring urgent attention: more than 65 million people worldwide are displaced (UNHCR, 2016); 9 of the 10 warmest years on record have occurred since 2000 (NASA, 2017); 1 person in 10 worldwide lacks access to safe drinking water (WHO, 2015). Addressing such complex issues requires engaging with various stakeholders (e.g. students, teachers, communities) across multiple perspectives (e.g., political, social, economic, cultural) and fields (e.g., STEM, humanities, arts). Teaching Mathematics for sustainable development is one approach to prepare students and teachers for both reading (understanding) and writing (transforming) the world with mathematics (Gutstein, 2006). This brings 21st century skills such as critical thinking, ethics and social responsibility (Ananiadou & Claro, 2009) to issues set within multidisciplinary STEM contexts. However, our literature search revealed that little is known about what is involved in teaching or learning to teach mathematics using big data sources to educate students on issues of Sustainable Development Goals (SDGs) and Sustainability. While the official and academic databases continue to play important roles for storing big data, it can be predicted that in the near future many new big data sources will be available at higher speeds (Karen et al. 2020). Schoolteachers need to be aware about the limits and opportunities in utilizing those emerging data sets in mathematics classroom contexts. The overarching goal of this project is to explore innovative ways to teach students the application of mathematical concepts in the use of big data to inquire about real-world issues in connection with Sustainable Development Goals (SDGs) and sustainability. The SDGs are universal goals for all people and are inherently global in nature. Learning about these initiatives through mathematics draws students’ attention, and helps develop insights into socio-economic and political issues within their local communities and around the world, such as lack of access to clean water, gender equality etc. (UNESCO, 2015). These issues are inseparable from “culture, and to truly understand the SDGs and other sustainability related issues, students need to learn about their communities and the world around them” (4 Reasons to Teach the Sustainable Development Goals, para. 4). We see integration of big data and mathematics concepts as a context for students to learn about their local communities and the world around them, explore and become agents of change on issues related to sustainable development goals and sustainability (beyond ecological and environmental).
• GPD-Cult17 - Viral Views

  What do we say about COVID-19 and how is this viewed? As we anticipate a second wave of the pandemic the social license for ongoing regulation is clearly at risk, especially among youth. Communications strategies have proven important in convincing Canadians to change their behavior. As we stay isolated for months, social media is also becoming a site for discussion and skepticism of regulations. Viral Views will adapt digital humanities methods to studying public discourse around the pandemic, returning visualizations to the public for reflecting on our discourse.

• NO-Eth19 - MOU with AI Global

  AI Global is a non-profit dedicated to advancing the responsible use of Artificial Intelligence (AI) through the development of practical tools that support business, researchers, government, civil society, and academia. Founded in 2017 by business and technology experts, AI Global is dedicated to bringing a diverse group of subject matter experts together to address regulatory and governance challenges facing the responsible implementation of AI. One of AI Global’s key initiatives is working with its partners on the creation of a Responsible AI certification mark (“the Check”) for consumers both in business and society to have more trust and confidence in the AI products, tools, and services they are using. Based on a robust and measurable AI Trust Index, the Check is continuously aligned with principles, best practices, and standards for responsible design, implementation, and use of AI. AI Global seeks to partner with University of Alberta on these initiatives as this work references, supports, and helps to advance related University of Alberta’s activities.

• NO-SciCom18 - The AI4Society Dialogues Podcast

  How will AI shape society and how will society shape AI? AI4Society Dialogues talks to leading researchers across the disciplines at the University of Alberta, a global leader in artificial intelligence research. The podcast explores how researchers are constructing and using AI in the course of their work, and examines opportunities, challenges, and concerns as AI becomes an increasingly prevalent aspect of our world.

  Season 1
  Partnering with the University of Alberta KULE Institute we present 10 episodes where we interview UofA reasearchers that use AI on their disciplines. Learn here about AI and: games, music, health, law, etc!

  Season 2
  This time we partner with our collegues of Precision Health Signature area. Listen to 10 new episodes, this time oriented specifically to AI and health.

• P-Apps26 - AI-Powered Solutions for Advanced Multi-Temporal Modelling and Optimisation of CO2 Transport, Utilisation and Storage Networks – ACTiON (ES+AI4Society)

  The overall objective of the proposed project is the application of AI-powered products to understand the relationship between subsurface systems utilised for geological storage, primarily CO2, but including competing demands of other streams such as H2, with the design and operation of a regional CO2 Capture, Utilisation and Storage (CCUS) Network. The focus is on tactical dynamic operability of the network in the short term, which has been largely ignored in earlier CCUS supply chain optimisation studies. Such models have been treating subsurface storage as a static box, defined by geological storage type, depth and number of injection wells, where the dynamics conditions of injectivity and storage capacity are ignored. In deep geological storage, the performance of individual wells is especially critical during injection because the operations are likely to be scheduled with a minimum number of injection wells for cost and risk assessment reasons. In several deep CO2 geological storage and CO2-EOR projects, the injectivity was found to be unexpectedly smaller than was anticipated during CO2 injection/floods due to complex interactions which should be included in dynamic network optimisation.

• P-Apps27 - Estimating Occupant Presence and Actions using Machine Learning Techniques (ES+AI4Society)

  Buildings account for a large portion of the total energy consumption in Canada. Most of this energy is consumed to maintain occupants' thermal and visual comfort. Thus, understanding the building occupancy is critical for designing energy-efficient building controls, maximizing human comfort, and creating a safe and healthy workplace during the pandemic. Despite the importance of incorporating occupancy information in the control of HVAC, lights, and window blinds, this information is often unavailable or inaccurate. Machine learning models can help fuse data from various sources (e.g., WiFi association logs, prox cards, cameras) to estimate occupancy at different spatial and temporal resolutions. Leveraging reinforcement learning algorithms, it is possible to determine optimal (energy-efficient) control policies which satisfy stringent safety requirements.

  This project aims to analyze the estimated building occupancy levels, whole-building energy consumption patterns, and operational data from the Building Management System (BMS) to 

  1. Refine the estimated number of occupants in each building and identify/address data quality issues. This is an important task as there is not a one-to-one mapping between WiFi users and building occupants. The refined occupancy information will be useful for monitoring building occupancy and issuing warnings if social distancing rules are violated (during the pandemic).
     
     
  2. Understand what percentage of the whole-facility energy use can be attributed to occupants, and utilizing this information to optimize the operation of HVAC and lighting systems. The optimal control policy is expected to reduce the building energy use while satisfying comfort requirements.
• P-Apps28 - The Improbable Persistence of low probability, high consequence events (ES+AI4Society)

  We are now applying these techniques to examine the improbable persistence of incidents in electricity generation and distribution. Significant improvements have been made in process and personal safety over the past decades. Nonfatal injuries and illnesses have dropped almost 70% since 1992. Yet, fatalities decreased at less than half that rate. Low probability, high consequence events have remained improbably persistent. Following an incident, an investigation will identify the causal and contributing factors, including organizational issues. However, this does not necessarily mean we can predict exactly how, when, or why the next accident or near miss will occur. This is particularly challenging for complex socio-technical organizations (i.e., energy, processing, transportation) with very low accident rates.

• P-Apps29 - The role of affect in online higher education: A comparative study in Canada and Mexico

  Several theories of learning and studies have shown the importance of affect in learning for a limited set of cultural contexts. This previous research has created a considerable body of work detailing the influence of how we feel (our affect) on our learning. However, we know very little about how students’ ability to monitor and regulate their affect (i.e., their meta-affective skills) relates to their learning or how this might differ across cultures. In online learning settings, we have the additional challenge of not being able to use sensors to monitor this latent aspect of student processes. We must instead rely on questionnaires that interrupt the learning activity.

  This project will begin by exploring differences in students’ meta-affective skills in two cultural settings, when student learning is mediated by online technologies. The first setting, Canada, has been reasonably well studied from the perspective of integrating and understanding student use of advanced educational technologies. The second, Mexico, has been largely ignored.

  After exploring potential differences across these contexts, we will proceed to integrating novel sensor technologies so that we can try to automatically detect both student affect and their meta-affective strategies when learning. Later work will consider how we can provide feedback to students that will encourage meta-affective strategy use to better support student learning in online settings.

• P-Apps30 - Construction Supply Chain Analysis on Forecasting the Demand for Small Equipment, Tools, and Consumables for Industrial Construction Projects This proposed research aims to improve the construction supply chain's efficiency, resilience, and agility by providing a data-driven forecast solution for small equipment, tools, and consumables throughout an industrial construction project's lifecycle. The proposed research is expected to provide insight into the historical demand for small equipment, tools, and consumables before and during the construction project. A variety of data modelling techniques (e.g., machine learning, statistical learning) will be used to develop an AI-enabled data-driven forecast solution on the demand of small equipment, tools, and consumables. Partnered with our industry organizations – PCL Industrial Management (general contractor) and Vallen Inc (industrial supplier), the proposed research is expected to develop solutions that can be leveraged among multiple stakeholders within the construction supply chain to improve transparency and predictability. Once applied in practice, the proposed solutions are expected to enhance the efficiency, productivity, agility, and capital allocation of the construction supply chain in small equipment, tools, and consumables, resulting in reduced waste, improved project performance and increased competitiveness.
• P-Apps31 - Tracing Conflict and Radicalization Online Malicious actors use social media to advance destructive agendas, including political extremism, racial hatred, and election disinformation. Armed groups use these platforms to recruit and radicalize Canadians and crowdsource funding. Meanwhile, right-wing extremists use social media to disseminate disinformation at a scale never before seen, transforming political discourse and undermining democracy. Political and military battles are now fought as much on social media platforms as on the frontlines. Yet, the extent and impact of malicious actors' disinformation and radicalization efforts on social media remain poorly understood. Social media platforms do far less to combat these problems than is commonly assumed. At the same time, social media data holds the potential to provide an unprecedented and expansive real-time record of online and real-world events and their impact on communities and individuals. For instance, social media has been used as source of evidence in many high-profile criminal investigations, such as the shooting down of the Malaysian Airlines MH17 flight over Ukraine in 2014 by Russian military. That investigation, like many others, followed the incipient science of Open Source Intelligence, whereby experts and volunteers from the public analyze text, audio, and video posted online to identify key actors and events underlying the conflict or investigation.
• P-Apps32 - Viewpoint Alberta Survey Databases

  Part of the Common Ground initiative, our Viewpoint Alberta team will be releasing one of the largest public opinion datasets in Alberta history, combining five (5) surveys collected over the course of the COVID-19 pandemic. 

  For more information on our project, please visit: https://www.commongroundpolitics.ca/viewpoint-alberta.

  Our project is led by Dr. Jared Wesley (Political Science), Dr. Michelle Maroto (Sociology), and Dr. Feodor Snagovsky (Political Science). It is funded by competitive grants from the Kule Institute for Advanced Study (KIAS), the Killam Trust, and AI4SOCIETY.

• P-Apps34 - Reinforcement Learning for Safe Handover between Humans and Autonomous Vehicles in Complex Environments

  Autonomous vehicles with various levels of high and full automation are revolutionizing Intelligent Transportation Systems (ITS) in terms of safety, energy consumption, and how humans interact with their cars. However, the driver intervention is required in highly dynamic environments (e.g., too many pedestrians or very busy traffic) or harsh weather conditions in which on board sensory information (from stereo/mono cameras, LiDARs, and radars) are not reliable for the safe decision making, motion planning, and controls in existing automated driving systems. During these safetycritical handover transitions between the human and autonomy, the vehicle knows little about the human’s capability/prediction, and the human is unaware of the vehicle’s handling capacities, thus, could take actions that deteriorate the stability and imperil the safety of passengers and pedestrians.

• P-Apps35 - Telegram Observatory The Telegram social media environment is used extensively by extreme groups to communicate. We propose to develop the infrastructure at the U of Alberta that can follow specific Telegram channels in order to study communities, especially Alt-Right communities in the province. Further we propose to add search and analysis tools with AI capabilities to study the social media archives added. This will add to the infrastructure available for researchers to study social media and allow us to use different AI tools to study social media.
• P-Cult33 - The Social Life of Cinema During COVID

  Using the Kinomatics Showtime database we are seeking to understand how cinemagoing - as a relationship between films, venue locations and audiences, was affected by the global spread of COVID-19. We have collected data that describes all movie showtimes and cinema locations for all films released in the majority of global jurisdictions between November 2019 and June 2021 inclusively. This period is particularly interesting because it covers the onset and diffusion of COVID-19 around the world. Our data includes approximately 4890 time-stamped XML files, with an average of 3860 datapoints in each file (around 19 million datapoints in total). Earlier baseline data collected between 2012 and 2015 will be used to measure the extent of longitudinal and seasonal change. This earlier data comprises almost 18 million datapoints.

• P-DHLTH24 - Early Detection of Depression and Other Mental Disorders Using Machine Learning

  Depression is one of the most common mental disorders that affects more than 320 million people worldwide.  Depression is considered a primary cause of disability and suicide. Regrettably, less than 50% of individuals suffering from depression receive proper treatment. Moreover, depression causes one death every 40 seconds, adding up to 800,000 deaths by suicide worldwide every year.

  The joint research between the University of Alberta and Tecnológico de Monterrey will develop data infrastructure and machine learning tools to detect depression, by analyzing conversational audio, biomarkers, and psychophysiological profiles. With all these data, a new and novel dataset will be created. This dataset will be the essentials for well designed predicting machine learning models. The joint team will develop the machine learning model(s) best suited to the dataset and determine with high accuracy, the depression level of a person. Additionally, it would be the first dataset of its kind that can be used by the research community for future machine learning studies on depression. Moreover, we envisage the development of a platform based on wearables and embedded artificial intelligence capable of alerting the subjects of psychophysiological symptoms related to depression in real-time. With this preliminary assessment, the individual can promptly be directed to the appropriate channel (i.e. a psychological service), where trained personnel assist in avoiding any negative impact that the depressive symptoms could have in his/her life and environment. 

   

• P-DHLTH25 - SMiLE

  Social Media and the Transformation of Healthcare: A Viewpoint

  We examine the complex intersection of social media and SDoH, a relationship informed by their impact on one another. We argue that a paradigm shift in the physician-patient relationship is warranted where physicians: a) acknowledge the impacts of Social Determinants of Health on information-seeking behavior, b) recognize the positive and negative roles of social media as a mediator of health through the lens of SDoH, and c) use social media to catalyze positive changes in the standard of care.

  We suggest that only by broadening our understanding on the intimate linkage between social media and SDoH the gap between current approaches to patient care and the vision of precision medicine with its inherent personalized aspects can be effectively bridged.

   

• P-DHLTH26 - Application of Face Recognition Methodologies for Antibody Selection for COVID-19

  The impact of the ongoing COVID-19 pandemic and the real possibility of future CoV-related epidemics both underscore the need for rapid development of new treatments and efficient intervention protocols. In this regard, protein–protein interactions (PPIs) play an important role in a wide range of biological processes. They are involved in the principal activities and pathways at the cellular level, making them promising targets for drug discovery. The SARS-CoV2 spike protein plays a key role in mediating virus–cell fusion and the integration of the SARS-CoV2 virus into host cells. It has been demonstrated that the human monoclonal antibodies can be used as immunoprophylaxis or therapy against coronaviruses such as SARS-CoV [1]and MERS-CoV [2]. In this respect, the repurposing of available antibodies or the discovery of novel ones that can strongly interact with COVID-19 spike proteins can substantially inhibit the virus from affecting human cells.

  Our proposed project aims to develop a platform to facilitate repurposing of available antibodies to treat COVID-19 through identification of a novel protein–protein interaction protocol. It should be noted that the proposed research project directly aligns with the mandate of the AI4Society, as it aims to “develop appropriate computational platforms to implement AI technologies in a real-world use case” and it has direct applications in a vertical sector described by the AI4Society (i.e., health).

• P-DHLTH36 - Wearable oximetry and a virtual ward to improve outcomes of infants with Bronchopulmonary Dysplasia Infants with bronchopulmonary dysplasia (BPD), are at risk of poor cardiorespiratory control, and chronic and intermittent hypoxemia for months after birth. They have up to 60 episodes of intermittent hypoxemia (IH) per day. IH leads to apathological cascade of oxidative stress and injury and clinical morbidities including cognitive and language delay, motorimpairment, and death. Up to 30% of hypoxemic spells in children are clinically undetectable. In the neonatal intensive care unit (NICU), pulse oximetry is used to monitor and adjust oxygen needs of preterm infants up to 41 times/day. This vigilance is crucial because untreated hypoxemia for even a few minutes causes pulmonary inflammation, neurological damage, and death. The current problem: At home, this vigilance is not possible. The oximeters used in NICUs are cumbersome and rapidly cause parental fatigue and incompliance . Even if an infant is monitored, difficulties intransmitting data for clinical evaluation prevents timely intervention. Lack of evidence leads to lack of consistent advice, variability in practice and parental anxiety, frustration, and non-compliance. Some parents resort to non-medical grade oximeters with questionable accuracy. Others (29%) become non-compliant, and 32% self- wean their infants from oxygen without medical guidance. The current outcome: By 1 year, 53% of infants with BPD would have been hospitalized at least once, mostly (81%) for respiratory problems. Many have >2 emergency department presentations, >16 specialist attendances, >33 pathology tests, and>6 diagnostic imaging tests, costing ~$182,312/year to health system and at least ~$2,212 to parents. Later, respiratory and neurological function declines, resulting in poor quality of life even death in adulthood. What must we change? Oxygen is one ofthe 479 World Health Organization (WHO) essential medications that are vital for human health. To ensure oxygen is given safely,the WHO Pulse Oximetry project recommends monitoring of every child at risk of hypoxia. Due to lack of high-levelevidence, guidance for care of Canadian infants with BPD are unclear, leading to the health problems that stretch alreadyscarce resources. Oxygen management is also inequitable, with remote and Indigenous infants being 3 times less likely to be prescribed home oxygen therapy. What we offer: We will use technological advances spurred by COVID-19 pandemic to digitally innovate oxygen needs of infants with BPD at home. Using a wearable oximeter and a virtual care package (Masimo Safety Net) that was used safely and effectively to manage thousands of adult COVID-19 patients at home instead of the hospital, we will aim to improve oxygen management and health and decrease hospitalization in BPD infants. This is a low-risk, COVID-Safe proposal using devices and communication systems (i.e., smartphones and Wi-Fi). The overall aim will be is a large RCT, which will bring equity of care to remote and Indigenous families with difficulties accessing medical care, fills a knowledge gaps by finally acquiring home oximetry data and future proof home management of infants at risk of hypoxia with the creation of predictive algorithms from our innovative data.
• P-DHLTH37 - Proteinscanner – Hybrid Approaches for Rapid Screening of Protein Surfaces Using Geometric Fingerprints

  Proteins are “social molecules.” Every protein in a living cell interacts with many other proteins and small-molecule ligands to carry out its biological function. The study of these protein interactions is fundamental for understanding biological processes for all organisms, in both healthy and diseased states. With advances in computational methods and in computer hardware resources, the ability to accurately and rapidly predict new protein interactions has emerged as an attractive approach and an excellent complement to experimental techniques. Methods have been developed by which we can study the interactions between proteins and other biomolecules by considering their overall sequence and structural similarities1,2. Nevertheless, it is well established that the overall sequence and fold similarities of proteins do not necessarily translate to similarities in protein function or to interactions between similar proteins3. It should be noted that the interactions of a protein with other proteins and with biomolecules primarily involve amino acids at the surface of a protein. This surface can be represented as a continuous surface that is imprinted with geometric and chemical features (“fingerprints”). In turn, it can be assumed that proteins engaging in similar interactions share common fingerprints on their surface representation4–6. Fingerprints may be difficult to grasp by visual analysis but could be learned from large-scale datasets using artificial intelligence.

• P-Eth21 - Access to Justice Apps: A Tool to Support Accountable Policing?

  PI Dr. Nidhe Hegde – Computing Science
  EPS Partner Dan Jones – Edmonton Police Service

  Description This project builds on pre-existing legal apps developed at UAlberta to support access to justice. It seeks to understand and test how such tools might - by earning trust - increase trust in a policing system that centres community members as beneficiaries. The access to justice apps help members of the public assert fairness in policing. They help people submit complaints related to police conduct, and independently verify their rights during live interactions with police. Aggregate data from these apps could inform practice and policy development, if appropriate controls for fairness and privacy are possible. These need to be explored with vulnerable population protection as a top priority. This research project will examine one or more access to justice apps to consider the extent to which the data, interface and content aspects of the apps protect user privacy and promote fairness when deployed. Next, the project will consider if and how aggregate data from such apps could be ethically collected and analysed in a privacy-preserving manner. We hope this baseline work lays a foundation for developing further partnerships that bring together law and computing science in the service of communities that are the intended beneficiaries of access to justice apps. With some baseline understanding of the apps and their potential, we hope to have a good foundation to respectfully engage communities directly.

• P-Eth22 - Human-in-the-loop software for improving latent fingerprint identification

  PI Dr. Matthew E. Taylor - Alberta Machine Intelligence Institute | Dr. Miguel A. Medina-Pérez - Technológico de Monterrey
  EPS Partner Devin Laforce – Edmonton Police Service Brad Mandrusiak – Edmonton Police Service

  This project has alse being supported via our ITESM-UofA Seed Grant Program

  _______________________

  Description Automated methods for identifying latent fingerprints (those which are not visible to the human eye) have grown in popularity in the past years. However, as the use of these systems grows, new challenges have emerged that impede the marking of some of the basic features needed to represent fingerprints. Even the absence of two features can make a fingerprint no longer identifiable, creating barriers to the implementation of automated methods. To meet this challenge, we are developing an AI-enabled system to improve feature labelling for latent fingerprint identification. The system will predict the impact of including or not including certain features for a latent fingerprint identification system. The project seeks to develop a tool that can be used to support trained fingerprint examiners and in the training of individuals with little or no experience in the field. The project will focus on human-in-the-loop, a method of AI development that requires human interaction with the system, and will be supported by the Edmonton Police Service.

• P-Eth23 - Building a Partnership to Catalyze Police System Evolution

  PI Dr. Melissa Tremblay – Educational Psychology
  EPS Partner Dan Jones – Edmonton Police Service

  Description Controversial calls for police reform are echoing across the world, and the voices of youth rarely reverberate in the polarizing debates on this topic. Thus, there are four objectives for this project. We will aim to (1) solidify an emerging partnership of service providers, youth, and academics who are interested in understanding how police can better work with marginalized youth, (2) conduct a realist synthesis of strategies for police to work effectively with marginalized youth, (3) create a robust research proposal based on our realist synthesis, and (4) document our partnership processes. We will build on relationships with the Edmonton Police Service and with two community organizations that serve Indigenous youth, sexual and gender minority youth, and youth experiencing homelessness, addictions, and mental health challenges. Our ultimate aim is to lay the groundwork for shaping actionable recommendations for police systems to improve their ability to serve marginalized youth.

• P-Eth39 - The Politics of Policy Communication This proposal outlines our request to incorporate AI4Society into our modest research project. The project is structured as an honours research project between Dr. Jared Wesley and Michael Griffiths, with the collaboration of Dr. Lianne Lefsrud and one of her graduate students. We are looking to weave AI into the project to improve the project’s capability and minimize researcher bias. We believe that, not only can we enhance this project using AI, but also that we can develop a framework and software that can be used by future researchers. We are seeking support from AI4Society to greatly enhance our own project and build a foundation for future research, including the eventual submission of a SSHRC Insight Development Grant to study political communication in Western Canada and the Western United States.
• P-Eth40 - Copyright, Data Mining and Artificial Intelligence: Promoting the Public Interest in Research and Innovation Like most Copyright laws around the world, the Canadian Copyright Act was designed at a time when big data and large- scale computational processing of data had not become mainstream. However, given that works that are subject to copyright protection are being used as part of data sets for text and data mining activities, including the training of machine learning models, the issue of whether copyright laws need to react to this development has arisen internationally and domestically. In a recent public communication, Innovation, Science and Economic Development Canada (ISED) called for insights into copyright and related issues surrounding TDM, including the use of copyrighted works as training data for machine learning models and the development of AI systems. As Canada and other countries grapple with whether and how to regulate the use of in-copyright works for text and data mining activities and as training data for machine learning models, it is important and timely to examine from an interdisciplinary perspective, the interlinkages between copyright, data mining and artificial intelligence and how to ensure that Canada preserves and promotes the public interest in research and innovation. This project seeks to do that. It seeks to bring together a team of legal and non-legal researchers to consider important practical and technical issues in text and data mining, machine learning and the development of AI systems that implicate copyright rules in Canada. The insights from this interdisciplinary project would then be used to analyse the current state of copyright law in Canada with a view to determining whether the use of copyrighted works for text data mining activities, the training of machine learning models and the development of AI without seeking the permission of the copyright owner is lawful or not. The project would also consider the significance of an ecosystem where text and data mining and the development of AI is not unduly constrained by copyright law for the promotion and democratisation of research and innovation.
• P-IoT38 - Smart Home Energy Management Systems for Solar PV-Integrated Homes using Internet of Things Residential applications of solar PV systems represent both an emerging trend and a promising solution for mitigating greenhouse gas emissions from the housing sector and reducing energy costs for homeowners. To achieve optimal energy performance and a high load match index, the electricity generated by the solar PV system should be used on-site as much as possible with minimal interaction with the utility grid. Therefore, novel technologies and tools for energy management of solar PV-integrated homes are urgently needed. Home Energy Management Systems (HEMS) have seen increasing popularity and adoption in recent years. However, existing HEMS lack a critical component for achieving a high load match index in homes integrated with solar PV and IoT (Internet of Things) appliances/devices, as these HEMS do not use energy generation forecasting to schedule household activities in a manner that optimizes use of the generated energy. In this context, in this research we will develop multi- temporal solar PV energy generation prediction models that employ machine learning and deep learning to accurately predict the daily, hourly, and minutely energy generation based on the location of the house, solar PV layout (i.e., orientation, tilt angle, etc.) and other system parameters (i.e., number of panels, capacity, efficiency, etc.) and weather prediction. Then, a framework will be developed to use the energy generation predictions for scheduling the household activities that consume a significant amount of energy (such as washing and drying, dishwashing, ironing, water heating, air conditioning, recharging the storage, and charging of electric vehicle) using IoT. The proposed framework will bridge a very critical gap in the energy management of solar PV-integrated energy-efficient homes (and/or NetZero Energy Homes), using IoT to help homeowners reduce energy costs and while mitigating the adverse effects of increased residential solar PV use on the utility grid (or smart grid).
• P-M20 - Student Pathways

  Course selection and academic advising are difficult and time-consuming tasks. Thus far, no one has developed a tool that can help advisors or students create a personalized plan for when a student should take different courses within an academic program. We will use de-identified historical data to identify successful course enrollment patterns based on individual student backgrounds. We will then use these patterns to recommend course paths to students to help them select the sequence of courses that will ensure their successful completion of their current academic program.

• P-M30 - Black Youth Mentorship and Leadership Program and Research Internship Experience

  In 2016, there were around 1.2 million Black people in Canada, including 100,245 Black youths ages 15 to 19. Alberta has the fastest growing population of Black people in Canada: in 2016, there were 57,815 Black people in Edmonton, and the number is growing. Black youths experience social, economic, and health inequities in Canada, and are less likely to attain a post-secondary qualification. Among Black boys aged 13 to 17 in 2006, approximately half had a postsecondary qualification in 2016, compared with 62% of other boys, and this has been declining since 2011. Young Black males were nearly twice as likely as other young males to not be employed, attending school, or completing job training. Attitudes of teachers towards Black youths, the lack of African-Canadian history and culture in the educational curriculum, systemic anti-Black racism, and the lack of mentors contribute to poor educational outcomes, which translate to poorer economic outcomes and less food and job security. Black-focused education can improve the economic outcomes of Black youths. Furthermore, strong evidence indicates mentorship is effective across behavioral, social, emotional, and academic domains of youth development. To address these challenges, the United Nations recommends Canada should “strengthen Afrocentric education curricula”. Our project addresses this recommendation. Proponents of Afrocentric educational programs assert that Black youths suffer from low self-esteem, which leads to low school performance, but that they can be motivated toward high achievement by studying their history. Research also shows that alienation and expulsion rates decrease and self-esteem and university attendance rates increase when Black students are taught by Black teachers.