profile_2.jpg

Aaron J.
Havens


Postdoctoral Researcher
Meta Fundamental AI Research (FAIR)

I’m a postdoctoral researcher at FAIR on the Generative Modeling Foundations team under Brian Karrer and Yaron Lipman. I’m developing scalable methods for generative modeling and principled reward-based tilting / finetuning in both continuous and discrete domains. I’m especially passionate about modeling physical processes that help advance Engineering and science, taking into account geometric structures and invariants.

Previously I received a PhD in ECE from University of Illinois Urbana-Champaign, where I worked on robust control theory $\cap$ machine learning under Prof. Bin Hu. I also have an MS in Aerospace Engineering from UIUC where I primarily worked on learning algorithms for autonomous field-robotics under Prof. Girish Chowdhary. I’ve been fortunate to have several industry research experiences at Meta FAIR NYC, Peferred Networks Tokyo, TuSimple (autonomous semi-trucks) and NASA Jet Propulsion Lab (JPL).

I’m orignally from Des Moines, Iowa 🌽 and earned my Bachelors in Mechanical Engineering at Iowa State University, Ames, Iowa.

selected publications

  1. ICML
    molecules.gif
    Adjoint Sampling: Highly Scalable Diffusion Samplers via Adjoint Matching
    Aaron J Havens, Benjamin Kurt Miller, Bing Yan, and 8 more authors
    In Forty-second International Conference on Machine Learning (also featured as Oral at Frontiers in Probabilistic Inference: Sampling Meets Learning. ICLR Workshop), 2025
  2. ICML
    Fine-grained local sensitivity analysis of standard dot-product self-attention
    Aaron J Havens, Alexandre Araujo, Huan Zhang, and 1 more author
    In Forty-first International Conference on Machine Learning, 2024
  3. ICLR
    On the scalability and memory efficiency of semidefinite programs for lipschitz constant estimation of neural networks
    Zi Wang, Bin Hu, Aaron Havens, and 4 more authors
    In , 2024
  4. ICLR
    Novel Quadratic Constraints for Extending LIPSDP Beyond Slope-restricted Activations
    Patricia Pauli, Aaron Havens, Alexandre Araujo, and 4 more authors
    In 12th International Conference on Learning Representations, 2024
  5. NeurIPS
    Exploiting connections between Lipschitz structures for certifiably robust deep equilibrium models
    Aaron Havens, Alexandre Araujo, Siddharth Garg, and 2 more authors
    Advances in Neural Information Processing Systems, 2023
  6. ICLR
    A unified algebraic perspective on lipschitz neural networks
    Alexandre Araujo, Aaron J Havens, Blaise Delattre, and 2 more authors
    In The Eleventh International Conference on Learning Representations (Spotlight), 2023
  7. L4DC
    fvin.png
    Forced variational integrator networks for prediction and control of mechanical systems
    Aaron Havens and Girish Chowdhary
    In Learning for Dynamics and Control, 2021
  8. NeurIPS
    mlah.png
    Online robust policy learning in the presence of unknown adversaries
    Aaron Havens, Zhanhong Jiang, and Soumik Sarkar
    Advances in neural information processing systems, 2018