Hai Victor Habi

I’m a Ph.D. student in the School of Electrical Engineering at Tel Aviv University, under the supervision of Professor Hagit Messer (Tel Aviv University) and Professor Yoram Bresler (University of Illinois Urbana-Champaign). My research lies in the intersection of statistical signal processing and deep learning, with a focus on generative modeling. Specifically, I am developing methodologies for obtaining estimation performance bounds from data.

I previously earned my M.Sc. in Electrical Engineering from Tel Aviv University under the supervision of Professor Hagit Messer in the Cellenmon Lab where I dealt with different aspects of environmental monitoring using machine learning and commercial microwave links.

In addition, I am a Principal Research Scientist at Arm. Before joining Arm, I headed the research and algorithm development group at Sony Semiconductor Israel, focusing on efficient deep learning inference techniques — including quantization, pruning, and low-rank approximation.

Selected Publications

Data Generation for Hardware-Friendly Post-Training Quantization

Lior Dikstein, Ariel Lapid, Arnon Netzer, and 1 more author

In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2025

Bib PDF

@inproceedings{dikstein2025data,
  title = {Data Generation for Hardware-Friendly Post-Training Quantization},
  author = {Dikstein, Lior and Lapid, Ariel and Netzer, Arnon and Habi, Hai Victor},
  booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  year = {2025}
}

Learned Bayesian Cram\backslash’er-Rao Bound for Unknown Measurement Models Using Score Neural Networks

Hai Victor Habi, Hagit Messer, and Yoram Bresler

arXiv preprint arXiv:2502.00724, 2025

Bib

@article{habi2025learned,
  title = {Learned Bayesian Cram$\backslash$'er-Rao Bound for Unknown Measurement Models Using Score Neural Networks},
  author = {Habi, Hai Victor and Messer, Hagit and Bresler, Yoram},
  journal = {arXiv preprint arXiv:2502.00724},
  year = {2025},
}

Learning the Barankin lower bound on doa estimation error

Hai Victor Habi, Hagit Messer, and Yoram Bresler

In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024

Bib

@inproceedings{habi2024learning,
  title = {Learning the Barankin lower bound on doa estimation error},
  author = {Habi, Hai Victor and Messer, Hagit and Bresler, Yoram},
  booktitle = {ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages = {9906--9910},
  year = {2024},
  organization = {IEEE}
}

Fully Quantized Neural Networks for Audio Source Separation

Elad Cohen, Hai Victor Habi, Reuven Peretz, and 1 more author

IEEE Open Journal of Signal Processing, 2024

Bib

@article{cohen2024fully,
  title = {Fully Quantized Neural Networks for Audio Source Separation},
  author = {Cohen, Elad and Habi, Hai Victor and Peretz, Reuven and Netzer, Arnon},
  journal = {IEEE Open Journal of Signal Processing},
  year = {2024},
  publisher = {IEEE}
}

Learning to bound: A generative Cramér-Rao bound

Hai Victor Habi, Hagit Messer, and Yoram Bresler

IEEE Transactions on Signal Processing, 2023

Bib HTML PDF Code

@article{habi2023learning,
  title = {Learning to bound: A generative Cram{\'e}r-Rao bound},
  author = {Habi, Hai Victor and Messer, Hagit and Bresler, Yoram},
  journal = {IEEE Transactions on Signal Processing},
  volume = {71},
  pages = {1216--1231},
  year = {2023},
  publisher = {IEEE}
}

Towards Fully Quantized Neural Networks For Speech Enhancement

Elad Cohen, Hai Victor Habi, and Arnon Netzer

In Interspeech 2023, 2023

DOI Bib

@inproceedings{cohen23_interspeech,
  title = {Towards Fully Quantized Neural Networks For Speech Enhancement},
  author = {Cohen, Elad and Habi, Hai Victor and Netzer, Arnon},
  year = {2023},
  booktitle = {Interspeech 2023},
  pages = {181--185},
  doi = {10.21437/Interspeech.2023-883},
  issn = {2958-1796}
}

Learned generative misspecified lower bound

Hai Victor Habi, Hagit Messer, and Yoram Bresler

In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023

Bib

@inproceedings{habi2023learned,
  title = {Learned generative misspecified lower bound},
  author = {Habi, Hai Victor and Messer, Hagit and Bresler, Yoram},
  booktitle = {ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages = {1--5},
  year = {2023},
  organization = {IEEE}
}

Hmq: Hardware friendly mixed precision quantization block for cnns

Hai Victor Habi, Roy H Jennings, and Arnon Netzer

2020

Bib HTML PDF Video

@book{habi2020hmq,
  title = {Hmq: Hardware friendly mixed precision quantization block for cnns},
  author = {Habi, Hai Victor and Jennings, Roy H and Netzer, Arnon},
  booktitle = {Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part XXVI 16},
  pages = {448--463},
  year = {2020},
  organization = {Springer},
}

Uncertainties in short commercial microwave links fading due to rain

Hai Victor Habi, and Hagit Messer

In ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2020

Bib HTML Video Code

@inproceedings{habi2020uncertainties,
  title = {Uncertainties in short commercial microwave links fading due to rain},
  author = {Habi, Hai Victor and Messer, Hagit},
  booktitle = {ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages = {9006--9010},
  year = {2020},
  organization = {IEEE},
}

Recurrent neural network for rain estimation using commercial microwave links

Hai Victor Habi, and Hagit Messer

IEEE Transactions on Geoscience and Remote Sensing, 2020

Bib HTML Code

@article{habi2020recurrent,
  title = {Recurrent neural network for rain estimation using commercial microwave links},
  author = {Habi, Hai Victor and Messer, Hagit},
  journal = {IEEE Transactions on Geoscience and Remote Sensing},
  volume = {59},
  number = {5},
  pages = {3672--3681},
  year = {2020},
  publisher = {IEEE},
}