Taija - Trustworthy AI in Julia

NWO ICT.Open 2023

Patrick Altmeyer

Delft University of Technology

May 14, 2024

🐶 Taija

Research informs development, development informs research.

Trustworthy Artificial Intelligence in Julia.

Trustworthy Artificial Intelligence in Julia.

Taija is a collection of open-source packages for Trustworthy AI in Julia. Our goal is to help researchers and practitioners assess the trustworthiness of predictive models.

Our work has been presented at JuliaCon 2022 and will be presented again at JuliaCon 2023 and hopefully beyond.

🛠️ Current State

Towards Trustworthy AI in Julia
  1. CounterfactualExplanations.jl (JuliaCon 2022)
  2. ConformalPrediction.jl (JuliaCon 2023)
  3. LaplaceRedudx.jl (JuliaCon 2022)
  4. AlgorithmicRecourseDynamics.jl (IEEE SaTML 2023)
  5. JointEnergyModels.jl (JuliaCon 2024 … I hope?)

… contributions welcome! 😊

🚀 Future Plans

  • Patrick has been actively developing Taija since starting his PhD in 2021 and plans to continue doing so until he graduates in 2025 and beyond.
  • TU Delft Student Software Project 2023: a total of 10 students will be working on Taija for the next 2-3 months.
  • We have managed to attract contributors from the Julia community and are looking for more.