Publications

Published Articles

• Workshop Affordable Winner's-Curse Correction for Single-GPU Hyperparameter Search
  O. Mudannayake
  GlobalSouthAI @ IJCAI-ECAI 2026, 2026
  [OpenReview: P8J9cLsmNs]

• Journal Privacy-Preserving ML via Secure Multi-Party Computing: Techniques and Trends
  O. Mudannayake, A. Indika, U. Jayasinghe, G. M. Lee, J. Alawatugoda
  Computers, Materials & Continua, 2025
  [doi: 10.32604/cmc.2025.068875]

• Conference Aya: Open-Access Multilingual Instruction-Tuning Dataset
  S. Singh, ..., O. Mudannayake, and others
  ACL 2024, 2024
  [doi: 10.18653/v1/2024.acl-long.620]

• Journal Detecting Water in UAV Image Streams with Flight Constraints
  H. Samaranayake, O. Mudannayake, D. Perera, P. Kumarasinghe, C. Suduwella, K. De Zoysa, P. Wimalaratne
  JVCI (Elsevier), 2023
  [doi: 10.1016/j.jvcir.2023.103933] [PDF]

• Journal ML and Deep Learning for Multi-Step Municipal Waste Forecasting
  O. Mudannayake^*, D. Rathnayake^*, J. D. Herath, D. K. Fernando, M. Fernando
  IEEE Access, 2022
  [doi: 10.1109/ACCESS.2022.3221941] [PDF]

• Conference kMatrix: Space-Efficient Streaming Graph Summarization
  O. Mudannayake, N. Ranasinghe
  IEEE ICIAfS, 2021
  [doi: 10.1109/ICIAfS52090.2021.9606137] [PDF]

^* Equal contribution

Articles Under Review

• Journal One Quarter of the World, Underserved: Deepfake Detection for South Asia
  D. Sewwandi, I. Malahat, I. Nanayakkara, S. Kuikel, O. Mudannayake, D. Gamage, S. Herath, Y. R. Shrestha
  Under review at ACM Computing Surveys, 2026

• Journal Deepfake.help: Designing Cultural Context Aware Media Literacy Intervention for AI Generated Content for Global South
  D. Sewwandi, D. Gamage, O. Mudannayake, Y.-C. Lee
  Under review at Proceedings of the ACM on Human-Computer Interaction (CSCW), 2026

Preprints

• Preprint ImageLab: Simplifying Image Processing Exploration
  S. Dissanayaka^*, O. Mudannayake^*, T. Halloluwa, C. De Silva
  Preprint, 2024
  [arXiv: 2401.03157]

^* Equal contribution

Manuscripts in Preparation

• Manuscript A Cross-Space Audit of Intrinsic Reward Search in Open-World Reinforcement Learning
  O. Mudannayake

• Manuscript Pixel-Domain Sensory Curricula are Null in Mean; Privileged-Channel Withdrawal Shows Exploratory Between-Seed Variance Compression on Craftax Double DQN
  O. Mudannayake