Project: Towards End-User-Centered Explainable Artificial Intelligence

Team: Weina Jin
Advisors: Ghassan Hamarneh (SFU), Xiaoxiao Li (UBC)

Description: Being able to explain its decisions to end-users in understandable ways is a necessity to deploy artificial intelligence (AI)-backed decision support systems in risk-sensitive domains such as healthcare. Yet, existing explainable AI (XAI) techniques are designed for technical users and disproportionately ignore end-users' high demand for AI explainability. To democratize AI and make AI explanations unbiased and accessible for end-users, I collaborated with end-users (including laypersons and physicians) with a participatory design process to discover end-users' requirements for XAI. Grounded in users' insights, I then developed the Clinical XAI Guidelines, the End-User-Centered Explainable AI Framework EUCA, revealed ill practices of XAI evaluation in the community, and proposed new end-user-centered XAI techniques. These efforts inform AI researchers of end-users perspectives, and facilitate their technical specification and development of end-user-centered XAI techniques that respect end-users' reasoning and decision process, follow human communication norms with explanations, and align with end-users' values and utility. 

View Project Presentation Slides

Winner of the 2023 CS Diversity Award.

Project: Eliciting the Needs of Informal Learners of Computational Skills for the Design of Self-Monitoring Interventions using Participatory Research

Team: Rimika Chaudhury (PhD, CS), Taha Liaqat (BSc, CS)

Advisors: Parmit Chilana (SFU), Philip Guo (UC San Diego)

Description: Millions of adults are turning to informal learning resources online to study programming, web development, machine learning (ML), and other computational skills but often face barriers in self-directing and sustaining their efforts. We draw upon a human-computer interaction (HCI) approach to understand the challenges faced by diverse and underrepresented informal learners and explore interactive solutions to empower these learners to become more self-directed and engaged. Our studies have brought to the fore the stigma and challenges these learners face in finding a community among power users and university-educated CS professionals. Based on these insights, we are designing new interventions to support learners in self-directing their efforts in a way that will be inclusive and impartial to learners of all backgrounds. Our insights complement efforts in industry and academia to achieve a diverse, skilled workforce that is well-equipped to upskill and reskill their computational strengths for today’s fast-paced, high-tech industry.

View Project Presentation Slides

Project: Python Quizzlo

Team: Mehreen Uzma, Anagha Molakalmur Anil Kumar

Advisors: Harpreet Dubb (SFU WiCS Co-President), Denise Siu (SFU WiCS Co-President), sumo Kindersley (SFU CS Technical Support Staff)

Description: Python Quizzlo, a beginner-friendly python workshop taught to female high school students during try/CATCH 2022, strives to encourage girls along their journey in STEM. Try/CATCH is an outreach program hosted by the Women in Computer Science (WiCS) student club at SFU, with the initiative and goal to empower women. We provide a platform to explore the technological world surrounded by other women they can aspire to look up to. Particularly, Python Quizzlo aims to cater to those new to computer science. We taught approximately sixty students the basics of Python - a demanding programming language used widely in the workforce. We are setting them up for success while providing an opportunity to explore and learn from their peers who are there to uplift each other. We received high commendations for our workshop and inspired students to take up computer science in the future.

View Project Presentation Slides

Project: Color Invariant Representation Learning for Unbiased Classification of Skin Lesions

Team: Arezou Pakzad, Kumar Abhishek

Advisor: Ghassan Hamarneh (SFU)

Description: While deep learning-based approaches have shown expert-level performance in dermatological diagnosis tasks, they have also been shown to exhibit biases toward certain demographic attributes, especially skin types (e.g., light versus dark), which raises a fairness issue that needs to be addressed. In this project, we propose Color Invariant Representation learning for unbiased Classification of skinLesions (CIRCLe) for mitigating the impact of skin type biases in automated skin disease diagnosis. CIRCLe learns features that are invariant to skin color variations in dermatological images, leading to a more accurate and unbiased diagnosis across all skin types. Additionally, we propose a new metric for measuring the fairness of machine learning models that, unlike other fairness metrics currently in use, supports multiple protected groups. Our results on a large-scale public dataset show that CIRCLe achieves a new state-of-the-art performance on diagnosis accuracy, more than doubling the current accuracy, while improving fairness by 27.3%. 

View Project Presentation Slides

Project: Atomic Red Team

Team: Aman Saxena

Advisors: Mohammad Tayebi, Magno Logan

Description: Our project, "Enhancing the Atomic Red Team with Advanced Threat Emulation Techniques," was a joint venture between SFU and Trend Micro to improve the effectiveness of cybersecurity. With the guidance of Magno Logan, a prominent industry expert, we aimed to enhance the Atomic Red Team product by integrating advanced threat emulation techniques. The Atomic Red Team is a test suite that measures the efficiency of an organization's security measures. We aimed to strengthen this framework by incorporating advanced techniques, resulting in a more comprehensive and realistic assessment of an organization's security posture. This is crucial because cybersecurity is now a crucial component of modern society, with many facets of our lives dependent on technology. Our project is contributing to social good by improving the efficacy of security measures, advancing the field of cybersecurity, and ultimately aiding in safeguarding individuals and organizations from cyber threats.

View Project Presentation Slides