---

Preprints

• Fast Convex Quadratic Optimization Solvers with Adaptive Sketching-based Preconditioners. [pdf] [Code]
  Jonathan Lacotte and Mert Pilanci, 2022.
  

  In preparation

Publications

• Globally optimal training of neural networks with threshold activation functions. [pdf]
  Tolga Ergen, Halil Gulluk, Jonathan Lacotte and Mert Pilanci.
  

  ICLR, 2023.

• Adaptive and Oblivious Randomized Subspace Methods for High-Dimensional Optimization: Sharp Analysis and Lower Bounds. [pdf]
  Jonathan Lacotte and Mert Pilanci.
  

  IEEE Transactions on Information Theory, 2022.

• The Hidden Convex Optimization Landscape of Two-Layer ReLU Neural Networks: an Exact Characterization of the Optimal Solutions. [pdf]
  Yifei Wang*, Jonathan Lacotte* and Mert Pilanci.
  (* denotes equal contributions)
  

  ICLR 2022. Oral presentation

• Newton-LESS: Sparsification without Trade-offs for the Sketched Newton Update. [pdf]
  Michal Derezinski, Jonathan Lacotte, Mert Pilanci and Michael Mahoney.
  

  NeuIPS 2021. Spotlight presentation

• Adaptive Newton Sketch: Linear-time Optimization with Quadratic Convergence and Effective Hessian Dimensionality. [pdf]
  Jonathan Lacotte, Yifei Wang and Mert Pilanci.
  

  ICML 2021.

• Effective Dimension Adaptive Sketching Methods for Faster Regularized Least-Squares Optimization. [pdf]
  Jonathan Lacotte and Mert Pilanci.
  

  NeurIPS 2020. Oral presentation.

• Optimal Iterative Sketching with the Subsampled Randomized Hadamard Transform. [pdf] [Code]
  Jonathan Lacotte, Sifan Liu, Edgar Dobriban, and Mert Pilanci.
  

  NeurIPS 2020.

• Optimal Randomized First-Order Methods for Least-Squares Problems. [pdf]
  Jonathan Lacotte and Mert Pilanci.
  

  ICML 2020.

• High-Dimensional Optimization in Adaptive Random Subspaces. [pdf][Code]
  Jonathan Lacotte, Mert Pilanci, and Marco Pavone.
  

  NeurIPS 2019.

• A Risk-Sensitive Finite-Time Reachability Approach for Safety of Stochastic Dynamic Systems. [pdf]
  Margaret Chapman, Jonathan Lacotte, Aviv Tamar, Donggun Lee, Kevin Smith, Victoria Cheng, Jaime Fisac, Susmit Jha, Marco Pavone and Claire Tomlin.
  

  ACC 2019.

• Risk-sensitive Generative Adversarial Imitation Learning. [pdf]
  Jonathan Lacotte, Mohammad Ghavamzadeh, Yinlam Chow, and Marco Pavone.
  

  AISTATS 2018.

• Risk-sensitive Inverse Reinforcement Learning via Semi-and Non-parametric Methods. [pdf] [Code]
  Sumeet Singh, Jonathan Lacotte, Anirudha Majumdar, and Marco Pavone.
  

  The International Journal of Robotics Research 2018.

Technical reports

• Faster Least Squares Optimization. [pdf]
  Jonathan Lacotte and Mert Pilanci, 2021.

---

Thesis

• Fast Convex Quadratic Optimization Solvers with Adaptive Sketching-based Preconditioners. [link]
  Jonathan Lacotte, 2021.
  Advisors: Mert Pilanci, Stephen Boyd and Mary Wootters.
  

  Stanford University.