Sandamali Yashodhara Wickramasinghe
Personal links
About
Overview
Sandamali Yashodhara Wickramasinghe is a dedicated PhD candidate at the Artificial Intelligence Research Centre CitAI, with a strong passion for advancing in the field of Explainable AI. She is under the expert guidance of Dr. Jacob Howe and Dr. Laure Daviaud, whose combined expertise in AI, theoretical computer science, and mathematics guides her research.
Her research is supported by an EPSRC-funded initiative, and the core essence of her PhD project can be summarized in its mission to verify AI systems' behavior by extracting automata through a learning process. This entails using automata, mathematical models, to decipher the behavior of AI systems, such as neural networks. By applying these algorithms to extract models from recurrent neural networks and other AI systems, her work significantly contributes to the development of safer and more transparent AI technologies.
Qualifications
- BSc (Honours) in Statistics, University of Colombo, Sri Lanka, January 2016 - March 2020
Employment
- Graduate Teaching Assistant, City, University of London, United Kingdom, February 2023 - present
- Data Scientist, PickMe (Digital Mobility Solutions Lanka (PVT) Ltd), Sri Lanka, November 2020 - February 2023
- Associate Data Scientist, VeracityAI, Sri Lanka, March - September 2020
- Intern - Data scientist, VeracityAI, Sri Lanka, August 2019 - February 2020
Languages
English (can read, write, speak, understand spoken, peer review) and Sinhala; Sinhalese (can read, write, speak, understand spoken, peer review)
Publications
Publications by category
Conference papers and proceedings (2)
- Wickramasinghe, S.Y., Howe, J.M. and Daviaud, L. Extracting Deterministic Finite Automata from RNNs via Hyperplane Partitioning and Learning. .doi:10.1007/978-3-032-15638-9_8
- Wickramasinghe, S.Y., Howe, J.M. and Daviaud, L. (2025). Extracting Weighted Finite Automata from RNNs via iterative partitioning and spectral learning. International Workshop on Artificial Intelligence and Formal Verification, Logic, Automata, and Synthesis 26 October, Bologna.
Journal article
- Wickramasinghe, A. and Basnayake, W. (2022). A Comparative Study on Methods to Handle Wavelet Edge Distortion. American Journal of Applied Mathematics and Statistics, 10(3), pp. 95-102. doi:10.12691/ajams-10-3-4
Professional activities
Event/conference
-
[AUA] Design Thinking for Urban Challenges: Big Data Approach. (Workshop) Seoul National University, South Korea (2020).
Description: Learning and imparting knowledge on the concept and characteristics of the 4th industrial revolution, and the basic principles of big data artificial intelligence, and also to experience the actual utilization measures in solving urban issues and smart city planning.