Dyah Adila

UW-Madison CS Dept. adila@wisc.edu.

prof_pic.jpeg

Hello! I am a PhD student at UW-Madison, advised by Fred Sala. This past Fall, I completed an internship with Amazon AWS AI research with Shuai Zhang led by Bernie Wang and George Karypis.

Research Interest I’m interested in building robust and reliable machine learning (ML) solutions, especially in settings where access to labeled data is limited. Recently, I’ve been focusing on robustifying foundation models (both multi-modal and language models). I have worked on weak supervision – mainly in improving weak sources coverage and mitigating their biases; and also on OOD detection – to understand the characteristics of OOD samples.

Originally, I was from Indonesia. I enjoy reading books (fantasy, thriller, mystery, theology, investigative journalism.. almost everything really!), drawing, doing pilates, and hiking.

See my CV

news

Feb 9, 2024 Our team (with Changho Shin) is shortlisted as one of the finalists in Qualcomm Innovation Fellowship 2024! Yeay lets gooo
Jan 16, 2024 Longer version of RoboShot is accepted to ICLR 2024!
Dec 17, 2023 Presented RoboShot, our work on robustifying Foundation Models without any extra labels and finetuning, as an oral oresentation in NeurIPS 2023 R0-FoMo workshop.

selected publications

  1. ICLR 2024
    Zero-Shot Robustification of Zero-Shot Models
    Adila, Dyah, Shin, Changho, Cai, Lingrong, and Sala, Frederic
    In ICLR 2024 May 2024
    PDF
  2. NeurIPS 2023
    Mitigating Source Bias for Fairer Weak Supervision
    Shin, Changho, Cromp, Sonia, Adila, Dyah, and Sala, Frederic
    In NeurIPS 2023 Mar 2023
    PDF
  3. NeurIPS 2023
    Geometry-Aware Adaptation for Pretrained Models
    Roberts, Nicholas, Li, Xintong, Adila, Dyah, Cromp, Sonia, Huang, Tzu-Heng, Zhao, Jitian, and Sala, Frederic
    In NeurIPS 2023 Oct 2022
    PDF
  4. NeurIPS
    AutoWS-Bench-101: Benchmarking Automated Weak Supervision with 100 Labels
    Roberts, Nicholas, Li, Xintong, Huang, Tzu-Heng, Adila, Dyah, Schoenberg, Spencer, Liu, Cheng-Yu, Pick, Lauren, Ma, Haotian, Albarghouthi, Aws, and Sala, Frederic
    In NeurIPS 2022 Jun 2022
    PDF
  5. UAI
    Shoring Up the Foundations: Fusing Model Embeddings and Weak Supervision
    Chen, Mayee F, Fu, Daniel Y, Adila, Dyah, Zhang, Michael, Sala, Frederic, Fatahalian, Kayvon, and Ré, Christopher
    In Proceedings of the conference on Uncertainty in artificial intelligence Aug 2022
    PDF
  6. NeurIPS
    Understanding Out-of-distribution: A Perspective of Data Dynamics
    Adila, Dyah, and Kang, Dongyeop
    In ICBINB@NeurIPS 2021 Nov 2021
    PDF