Bio

As a senior data scientist at Global Relay and the founder and research director of DG Sculptor, a nonprofit focused on AI and machine learning research, I have a strong background in the field. Before these roles, I served as the director of research and development at Humanitas Solutions, leading various R&D projects. I received my Ph.D. from McGill University and spent an additional year as a research associate at the Statistical Signal Processing Lab. My research interests currently include advanced AI/ML-based multi-media signal processing, including areas such as speech synthesis and implicit neural representation for AR/VR

About

As a researcher, engineer, and problem-solver, I am skilled in advanced signal processing, machine learning, and various areas of applied mathematics, including matrix analysis, optimization theory, modern optimization, probability, and stochastic processes. I am also an out-of-the-box thinker and a skilled debugger.
However, my greatest strength is my approach to problem-solving. As a fundamentalist scientist, I first understand the problem at hand and then seek out the necessary tools to solve it, rather than relying solely on my skills. I believe in following the scientific process, as advocated by Abraham Maslow's quote: "I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail." This approach allows me to be adaptable and efficient in finding solutions to complex problems.

Patents

Efficient Implementation of Hybrid Beamforming
U.S. Patent, US10790889B2
Alireza Morsali, Benoit JF Champagne, Afshin Haghighat
Efficient Receiver Combining for Hybrid Analog/Digital Beamforming
U.S. Patent, 20/024318 (Pending)
Alireza Morsali, Benoit JF Champagne, Afshin Haghighat

Publications

Resource-Constrained Data-Based Transfer Learning
IEEE Transactions on Neural Networks and Learning Systems (revision submitted)
Mohammad Askarizadeh, Alireza Morsali, Reza Rezaeian, Kim Khoa Nguyen
MLP-Attention: Improving Transformer Architecture with MLP Attention Weights
ICLR 2023
Alireza Morsali, Moein Heidari, Tohid Abedini, Samin Heydarian
DiffGANPaint: Fast Inpainting Using Denoising Diffusion GANs
ICLR 2023
Moein Heidari, Alireza Morsali, Tohid Abedini, Samin Heydarian
Difference Convex (DC) Programming Approach as an Alternative Optimizer for Neural Networks
IEEE ICC 2023 (Accepted)
Mohammad Askarizadeh, Alireza Morsali, Mostafa Zangiabadi, and Kim Khoa Nguyen.
Modeling and optimizing resource-constrained instance-based transfer learning
IEEE GLOBECOM 2022
Mohammad Askarizadeh, Mustafa Hussien, Alireza Morsali, K. Khoa Nguyen
Light-SERNet: A lightweight fully convolutional neural network for speech emotion recognition
ICASSP 2022
Arya Aftab, Alireza Morsali, Shahrokh Ghaemmagham, Benoit Champagne
Multi-Head ReLU Implicit Neural Representation Networks
ICASSP 2022
Arya Aftab, Alireza Morsali, Shahrokh Ghaemmagham
Deep Learning-Based Hybrid Analog-Digital Signal Processing in mmWave Massive-MIMO Systems
IEEE Access 10, 72348-72362
Alireza Morsali, Afshin Haghighat, Benoit Champagne
Ensemble Neural Representation Networks
Neural Processing Letters (submitted)
Milad Soltany Kadarvish, Hesam Mojtahedi, Hossein Entezari Zarch, Amirhossein Kazerouni, Alireza Morsali, Azra Abtahi, Farokh Marvasti
All-Analog Structures for AF Relaying in mmWave Massive MIMO Systems
ICASSP 2022
Soumya Khare, Alireza Morsali, Benoit Champagne
Generalized framework for hybrid analog/digital signal processing in massive and ultra-massive-MIMO systems
IEEE Access 8, 100262 - 100279
Alireza Morsali, Afshin Haghighat, Benoit Champagne
Coding advantage optimization of space-time-frequency block codes
Journal of Communications and Networks (Volume: 22, Issue: 2, April 2020)
Vahid Momenaei Kermani, Hossein Momenaee Kermani, Alireza Morsali
Achieving fully-digital performance by hybrid analog/digital beamforming in wide-band massive-MIMO systems
ICASSP 2020
Alireza Morsali, Benoit Champagne
Energy-Efficient Group-Sparse Transceiver Design for Multiuser MIMO Relaying in C-RAN
IEEE Transactions on Green Communications and Networking, (Volume: 4, Issue: 3, September 2020)
Jiaxin Yang, Ayoub Saab, Alireza Morsali, Benoit Champagne, Ioannis Psaromiligkos
Optimizing transmission rate in NOMA via block diagonalization beamforming and power allocation
IEEE PACRIM, 2019
Sara Norouzi, Alireza Morsali, Benoit Champagne
Design criteria for omnidirectional STBC in massive MIMO systems
IEEE Wireless Communications Letters (Volume: 8, Issue: 5, October 2019)
Alireza Morsali, Seyyed Saleh Hosseini, Benoit Champagne, Xiao-Wen Chang
Robust hybrid analog/digital beamforming for uplink massive-MIMO with imperfect CSI
IEEE WCNC 2019
Alireza Morsali, Benoit Champagne
Quasi‐orthogonal space–frequency and space–time–frequency block codes with modified performance and simplified decoder
IET Communications, Volume 11, Issue 11, 2017
Farokh Koroupi, Alireza Morsali, Vida Niktab, Mostafa Shahabinejad, Siamak Talebi
Realizing fully digital precoders in hybrid A/D architecture with minimum number of RF chains
IEEE Communications Letters (Volume: 21, Issue: 10, October 2017)
Alireza Morsali, Afshin Haghighat, Benoit Champagne
A new design criterion for linear receiver stbcs based on full-rank spaces
IEEE Communications Letters (Volume: 19, Issue: 2, February 2015)
Sadegh Tofigh, Hossein Momenaee Kermani, Alireza Morsali
Delay–interleaved cooperative relay networks
IEEE Communications Letters ( Volume: 18, Issue: 12, December 2014)
Mohammad Samavat, Alireza Morsali, Siamak Talebi
Coding advantage decomposition inequality for the space–frequency block codes
IET Communications, Volume 8, Issue 4 p. 500-507, 2014
Alireza Morsali, Sadegh Tofigh, Zahra Mohammadian, Siamak Talebi
On the coding advantages of the quasi‐orthogonal space–frequency block codes
IET Communications, Volume 8, Issue 4 p. 525-529, 2014
Mostafa Shahabinejad, Alireza Morsali, Siamak Talebi, Mohammad Shahabinejad
On permutation of space‐time‐frequency block codings
IET Communications, Volume 8, Issue 3 p. 315-323, 2014
Alireza Morsali, Siamak Talebi
Field extension and designing a new 2× 2 full-diversity, full-rate and information lossless space-time block code
FACTA UNIVERSITATIS (NIˇS) Ser. Math. Inform. 28, No 1, 2013)
Sadegh Tofigh, Alireza Morsali, Mohammad Askarizade, Hossein Momenaee Kermani
Texture-orientation based vanishing point detection using genetic algorithm
IEEE IST 2012
Mohammadhossein Barat, Saeid Saryazdi, Alireza Morsali, Hossein Nezamabadi-pour
The enhancement of system performance in quasi-orthogonal space-time-frequency codes
IEEE SoftCOM 2011
Mostafa Shahabinejad, Alireza Morsali