Presenting CMU-Africa research around the world

Patricia Raffaele

Jul 15, 2026

Ntung Ngela Landon jumping in the air in front of buildings in Japan

Source: Ntung Ngela Landon

Landon enjoyed sightseeing in Japan after presenting research at the IEEE Cryptography, Security and Privacy Conference.

Research on bias in AI facial recognition systems led research assistant and recent Carnegie Mellon University Africa alumna Ntung Ngela Landon (MSIT ’26) to the IEEE Cryptography, Security, and Privacy Conference in Sapporo, Japan to present her recent research on the topic.

Landon and fellow alumna Floride Tuyisenge (MS ECE ’26) worked under the direction of CMU-Africa assistant teaching professor Jema David Ndibwile. Landon, Tuyisenge, and Ndibwile authored the paper, Architectural Bias in Face Presentation Attack Detection: A Comparative Study of Vision Transformers and Convolutional Neural Networks, which was presented and is now published as part of the conference proceedings.

“This second phase of the research builds on our initial work, which focused on the question of how we fix demographic bias in digital platforms,” Ndibwile said. The initial research looked at Presentation Attack Detection (PAD) systems, a critical security layer in biometric authentication, and explored how fairness can be integrated into the security layer without sacrificing performance.

The research team used a deep learning model to eliminate bias in facial recognition used in applications such as mobile phone authentications, online services, banking, and other applications. At the end of that research, the team was left with a question of which AI architecture would best complement the PAD, further supporting technology that operates without demographic bias.

“Now we are looking at if the AI architecture itself can reduce bias,” Ndibwile said. They advanced the research from using just the one local lightweight model to comparing how two state-of-the-art AI models respond to one data set limited to a few faces from Asia and Africa.

“We went with one model over the other because it captures skin tones from different ethnic groups much better than the other model, showing that it allows for less bias,” Landon said.

  • Ntung Ngela Landon with her arms spread. She's in front of a powerpoint presentation.
    Landon was the youngest person to present research at the IEEE Cryptography, Security and Privacy Conference held in Sapporo, Japan, in April.
  • A group of about ten people all holding framed documents
    As one of two presenters from Africa, Landon was proud to represent CMU-Africa at the IEEE Cryptography, Security and Privacy Conference.

    • At the conference, Landon had the opportunity to meet CMU-Africa alumnus and fellow mentee of Ndibwile, Patrick Mutabazi (MSIT ’23). Mutabazi credits Ndibwile with guiding him to earn his Ph.D. and helping him secure a scholarship at the Nara Institute of Science and Technology in Ikoma, Japan, where Ndibwile received his Ph.D.

    “It was wonderful to have a CMU-Africa connection in Japan. He made sure I was comfortable and had a good time. He had meals with me and took me on tours. It was quite easy for me to experience Japan because he was there.” Landon said. “It was a full circle moment for Patrick. Professor Jema took care of him as a student, and then he was able to take care of me as a student during this conference. I was nervous about my presentation. It was my first research presentation ever and having Patrick there helped me focus.”

    “Presenting helped me develop confidence that our research in the area of facial biometric security is relevant,” Landon said. As the youngest presenter at the conferencem and one of only two presenters from Africa, she was proud to represent CMU-Africa and her continent.

    As a result of the presentation, the research team is exploring collaborating with both a university in Brazil and a university in San Jose, California.

    As far as next steps for the research, Ndibwile said, “In the near future we can implement this model in an actual facial recognition system and ensure that it will respond with unbiased results whatever the environment happens to be.” He also plans to apply for a patent once the research concludes.

    I am really grateful that I have the opportunity to work with Professor Jema. I am appreciative of CMU-Africa because being part of this institution is the foundation of everything.

    Ntung Ngela Landon, Research Assistant, CMU-Africa

    Landon plans to continue her work with the research team for another year and looks forward to applying to Ph.D. programs.

    “I am really grateful that I have the opportunity to work with Professor Jema. I am also appreciative of CMU-Africa,” Landon said. “Because being part of this institution is the foundation for everything.”