Suryabhan Singh Hada AI Engineer LinkedIn San Francisco Email: suryabhan90 [at] gmail.com Social: [LinkedIn] Google Scholar

About

I am an AI Engineer at LinkedIn, where my work involves the design of Large Language Models, focusing on the development of user embeddings for various query searches across the entire LinkedIn platform. I earned my Ph.D. at UC Merced, advised by Miguel Á. Carreira-Perpiñán, where I primarily researched the interpretability of deep neural networks and their visualization. You can find my CV [here].

Ph.D. thesis

• Approaches to Interpret Deep Neural Networks
  University of California, Merced, USA, 2022.
  [external link] [slides]

Publications

• Carreira-Perpiñán, M. Á. and Hada, S. S. (2023):
  "Inverse classification with logistic and softmax classifiers: efficient optimization".
  Unpublished manuscript, 2023, arXiv:2309.08945.
  [external link] [paper preprint] [Matlab implementation]

  • Short version at the Workshop on Beyond first order methods in machine learning systems (ICML 2020)
    [external link] [paper preprint]

• Carreira-Perpiñán, M. Á. and Hada, S. S. (2023):
  "Very fast, approximate counterfactual explanations for decision forests".
  37th AAAI Conf. Artificial Intelligence (AAAI 2023), pp. 6935-6943.
  [external link] [paper preprint] [slides] [poster] [Python implementation (coming soon)]

  • Longer version: Mar. 5, 2022, arXiv:2303.02883
    [external link] [paper preprint]

• Hada, S. S., Carreira-Perpiñán, M. Á. and Zharmagambetov, A. (2023):
  "Sparse oblique decision trees: a tool to understand and manipulate neural net features".
  Data Mining and Knowledge Discovery,
  [external link] [paper preprint]
  Many of the figures in the publisher's version are badly messed up, with wrong labels. The arXiv paper has the correct figures.

  • Also as: Jan. 30, 2023, arXiv:2104.02922
    [external link] [paper preprint]

• Hada, S. S. and Carreira-Perpiñán, M. Á. (2022):
  "Sparse oblique decision trees: a tool to interpret natural language processing datasets".
  International Joint Conf. on Neural Networks (IJCNN 2022).
  [external link] [paper preprint] [slides] [© IEEE]

• Hada, S. S. and Carreira-Perpiñán, M. Á. (2022):
  "Interpretable image classification using sparse oblique decision trees".
  IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP 2022), pp. 2759-2763.
  [external link] [paper preprint] [slides] [© IEEE]

• Hada, S. S. and Carreira-Perpiñán, M. Á. (2021):
  "Exploring counterfactual explanations for classification and regression trees".
  Int. Workshop and Tutorial on eXplainable Knowledge Discovery in Data Mining (ECML 2021).
  [external link] [paper preprint]

• Hada, S. S. and Carreira-Perpiñán, M. Á. (2021):
  "Understanding and Manipulating Neural Net Features Using Sparse Oblique Classification Trees".
  28th IEEE International Conference on Image Processing (IEEE - ICIP 2021).
  [external link] [paper preprint] [© IEEE]

• Hada, S. S. and Carreira-Perpiñán, M. Á. and Zharmagambetov, A. (2021):
  "Sparse Oblique Decision Trees:A Tool to Understand and Manipulate Neural Net Features".
  Unpublished manuscript, April 6, 2021, arXiv:2104.02922
  [external link] [paper preprint]

• Carreira-Perpiñán, M. Á. and Hada, S. S. (2021):
  "Counterfactual Explanations for Oblique Decision Trees: Exact, Efficient Algorithms".
  35th AAAI Conference on Artificial Intelligence (AAAI 2021).
  [external link] [paper preprint] [Python implementation (coming soon)]

  • Longer version:
    Carreira-Perpiñán, M. Á. and Hada, S. S.
    "Counterfactual Explanations for Oblique Decision Trees: Exact, Efficient Algorithms".
    Unpublished manuscript, March 1, 2021, arXiv:2103.01096.
    [external link] [paper preprint]

• Hada, S. S. and Carreira-Perpiñán, M. Á. (2021):
  "Style Transfer by Rigid Alignment in Neural Net Feature Space".
  IEEE Conf. Winter Conference of Applications on Computer Vision (WACV 2021)
  [external link] [paper preprint] [supplementary material] [slides] [animations] [Python implementation (coming soon)] [© IEEE]

  • Longer version:
    Hada, S. S. and Carreira-Perpiñán, M. Á.
    "Style Transfer by Rigid Alignment in Neural Net Feature Space".
    Unpublished manuscript, Sept 27, 2019, arXiv:1909.13690.
    [external link] [paper preprint]

• Zharmagambetov, A.* and Hada, S. S.* and Carreira-Perpiñán, M. Á. and Gabidolla, M. (2021) (* means equal contribution):
  "Non-Greedy Algorithms for Decision Tree Optimization: An Experimental Comparison".
  International Joint Conference on Neural Networks (IJCNN 2021).
  [external link] [paper preprint] [© IEEE]

  • Longer version:
    Zharmagambetov, A.* and Hada, S. S.* and Carreira-Perpiñán, M. Á. and Gabidolla, M. (2019):
    "An Experimental Comparison of Old and New Decision Tree Algorithms".
    Unpublished manuscript, Nov. 8, 2019, arXiv:1911.03054
    [external link] [paper preprint]

• Carreira-Perpiñán, M. Á. and Hada, S. S. (2020):
  "Inverse classification with logistic and softmax classifiers: efficient optimization".
  Unpublished manuscript, 2020, arXiv:.
  [external link] [paper preprint] [Matlab implementation]

  • Short version at the Workshop on Beyond first order methods in machine learning systems (ICML 2020)
    [external link] [paper preprint]

• Hada, S. S. and Carreira-Perpiñán, M. Á. (2021):
  "Sampling the "Inverse Set" of a Neuron: An Approach to Understanding Neural Nets".
  28th IEEE International Conference on Image Processing (IEEE - ICIP 2021).
  [external link] [paper preprint] [slides] [poster] [© IEEE]

  • Longer version:
    Hada, S. S. and Carreira-Perpiñán, M. Á.
    "Sampling the "Inverse Set" of a Neuron: An Approach to Understanding Neural Nets".
    Unpublished manuscript, Sept 27, 2019, arXiv:1910.04857.
    [external link] [paper preprint] [animations]

  • Extended abstract at the Bay Area Machine Learning Symposium, Oct. 19, 2017 (BayLearn 2017)
    [external link] [paper preprint] [poster]
    

• Suryabhan Singh Hada (2014): "A Fast Encoding Method for Fractal Image Compression".
  International Journal of Computer Applications (IJCA), August 2014.,
  [external link]

UC Merced | My Home Page