Thomas Flynn

I am a PhD student in the CS department of the CUNY Graduate Center and member of the CoSSMO Lab directed by Prof. Felisa Vázquez-Abad. My research interests include Neural Networks, Optimization, and Computer Vision.


Office: Hunter North 1000C (CoSSMO Lab)


  • Gradient descent using duality structures
    T. Flynn
    preprint. pdf
  • Forward sensitivity analysis for contracting stochastic systems
    T. Flynn
    preprint. pdf
  • Data driven stochastic approximation for change detection
    T. Flynn, O. Hadjiliadis, I. Stamos, and F. J. Vázquez-Abad
    To appear in Proceedings of the 2017 Winter Simulation Conference, Las Vegas, Dec. 3-6, 2017. pdf
  • Convergence of one-step adjoint methods
    T. Flynn
    22nd International Symposium on Mathematical Theory of Networks and Systems, Minneapolis, USA, July 12-15, 2016. pdf
  • Online Classification in 3D Urban Datasets Based on Hierarchical Detection
    T. Flynn, O. Hadjiliadis, and I. Stamos
    International Conference on 3D Vision (3DV), Lyon, Oct. 19-22, 2015. pdf
  • Time-scale separation in recurrent neural networks
    T. Flynn
    Neural Computation 2015 27:6, 1321-1344. pdf
  • Online algorithms for classification of urban objects in 3D point clouds
    I. Stamos, O. Hadjiliadis, H. Zhang, and T. Flynn
    The second 3DIMPVT Conference, Zürich, Switzerland, Oct. 13-15, 2012. pdf


  • manseg is a program for manual labeling of 3D point cloud data. Think MS paint for point clouds. It can be used for generating ground truth data and for comparing two labelings of a 3D scene (for instance comparing ground truth data with inferred labels) bitbucket page