CS SEMINAR:Towards the Era of Explainable AI

Guest: Seda Polat Erdeniz

Title:Towards the Era of Explainable AI

Time: March 28, 2024, 15:00 

Location: FASS 1001

Abstract: Explainable Artificial Intelligence (XAI) is an emerging technology dedicated to elucidating the outcomes produced by machine learning (ML) models for users. Given that ML models often operate as "black box," users are left unaware of the processes behind specific results. XAI techniques are designed to enhance the transparency and trustworthiness of these ML models for both end-users and developers. On one hand,end-users can place trust in ML results accompanied by explanations generated through XAI methods and can be more convinced in employing those ML results in their decision making. On the other hand, developers of ML models can use theseexplanations to debug and refine models as needed. Notably, the European Union hasrecently enacted the AI-Act, mandating that all AI application results must be explainable to foster trust. Consequently, current XAI methods may prove insufficient to address the breadth of AI applications, necessitating a heightened focus on XAIresearch. This seminar will present state of the art XAI methodologies and explore potential research directions for future endeavors.

Bio: Seda Polat Erdeniz has a PhD (2019) degree in Computer Science with a thesis on Recommendation and Configuration Systems from Graz University of Technology, Austria and previously graduated with BSc (2008) and MSc (2010) degrees in Computer Science from Bogazici University, Turkey. She has 15+ years professional experience in IT projects in several roles such as: software developer, product owner, founder, researcher,and data scientist. Regarding the domain experience, eight years of expertise in smartcard-based cryptographic applications and (after the career change by 2016) eight years of expertise in AI and Data Science.