Abbas Mammadov

prof_pic.jpg

I am a final year CS & Math BSc. student at Korea Advanced Institute of Science and Technology (KAIST). I am a research intern at BISPL Lab under the supervision of Professor Jong Chul Ye and completed research internship at KAI Inc.. My research interests include, but are not limited to scaling the diffusion models to arbitrary geometries, advancing the applicability in inverse problems, building novel theoretically well-supported generative framework to learn probability distributions. I also researched on theoretical DL to investigate the data geometries and topologies. Recently, I have been collaborating with Professor Anima Anandkumar at Caltech, researching on the generalization of diffusion models in function spaces.

Furthermore, I enjoy solving and proposing competitive math problems, please see Math Olympiads for more details on my Olympiad background and contributions.

Download my CV and Transcript.


Interests
  • Deep Learning
  • Computational Imaging
  • Diffusion Models
  • Geometric DL
  • Inverse Problems
  • Competitive Math

News

Mar 06, 2025 Our Amortized Posterior Sampling (APS) paper has been accepted to ICLR 2025 FPI workshop
Mar 02, 2025 I moved to Pasadena, CA, US and started Research Assistant position at Caltech
Feb 24, 2025 I graduated from KAIST with Summa Cum Laude (highest) and Excellent Academic Records
Oct 10, 2024 Our Function Space Inverse Solver paper is accepted to NeurIPS ML4PS workshop
Jul 23, 2024 Served as a Leader at International Math Olympiad (IMO) 2024 and IMSC 2024

Publications

  1. ICLRW 2025
    amortized.png
    Amortized Posterior Sampling with Diffusion Prior Distillation
    Abbas Mammadov*, Hyungjin Chung*, and Jong Chul Ye
    The Frontiers in Probabilistic Inference: Sampling meets Learning (FPI) workshop at ICLR, Apr 2025
  2. NeurIPSW 2024
    F_diff_inv.png
    Diffusion-Based Inverse Solver on Function Spaces With Applications to PDEs
    Abbas Mammadov, Julius Berner, Kamyar Azizzadenesheli , and 2 more authors
    Machine Learning and the Physical Sciences Workshop at NeurIPS, 2024
  3. preprint
    Geometric Diffusion Models for Data Over Arbitrary Manifolds
    Byeongsu Sim*Abbas Mammadov*, Moo K. Chung , and 2 more authors
    preprint, 2024
  4. ICML 2024
    data_top.png
    Defining Neural Network Architecture through Polytope Structures of Dataset
    Sangmin Lee, Abbas Mammadov, and Jong Chul Ye
    International Conference on Machine Learning (ICML) SPOTLIGHT, May 2024
  5. KSC 2023
    artificial_barber.png
    Artificial Barber: Hair Color and Style transfer using GANs
    Abbas Mammadov, and Kaleb Mesfin Asfaw
    Korea Software Congress (KSC) ORAL, Nov 2023