Imane Hamzaoui

Hello, I'm Imane (The e is silent).

I'm a PhD candidate at Mila Quebec AI Institute and Université de Montréal. I'm part of the SNAIL Lab under the supervision of Pr. Shahab Bakhtiari. I was previously a research assistant at Mila. I have obtained a Masters and Engineering degrees in Computer Science specialized in Intelligent Systems and Data from Ecole Nationale Supérieure d'Informatique, Algiers.

I was a research assistant and intern at New York University Adu Dhabi where I worked on exploring the potential of different species' brain-inspired neural networks.

My research interests revolve around NeuroAI, World Models and Deep Learning in general.

News

Jan 21, 2025	Poster accepted at the Mila Annual Conference 2026
Aug 3, 2025	Paper "The Human Labour of Data Work: Capturing Cultural Diversity through World Wide Dishes" accepted at ACM CSCW 2025
Jul 11, 2025	Paper "Neural Architecture Search with Mixed Bio-Inspired Learning Rules" accepted at ECAI 2025
Apr 11, 2025	Paper "The World Wide recipe: A community-centred framework for fine-grained data collection and regional bias operationalisation" accepted at ACM FAccT 2025
Feb 16, 2024	Paper "Medical Neural Architecture Search: Survey and Taxonomy" accepted at IJCAI 2024
Feb 16, 2024	Paper accepted as a Tiny Paper @ ICLR 2024 🎉
Nov 05, 2023	Paper accepted at the IBM/IEEE AI Compute Symposium 2023 poster session
Oct 22, 2023	Check our latest SNN poster with School Of AI Algiers
Oct 01, 2023	I became the Women Techmakers Algiers lead

Recent publications

Analog In-Memory Computing with Uncertainty Quantification for Efficient Edge-based Medical Imaging Segmentation

I. Hamzaoui H. Benmeziane Z. Cherif K. Elmaghraoui

ICLR Tiny Papers 2024

Abs PDF

This work investigates the role of the emerging Analog In-memory computing (AIMC) paradigm in enabling Medical AI analysis and improving the certainty of these models at the edge. It contrasts AIMC's efficiency with traditional digital computing's limitations in power, speed, and scalability. Our comprehensive evaluation focuses on brain tumor analysis, spleen segmentation, and nuclei detection. The study highlights the superior robustness of isotropic architectures, which exhibit a minimal accuracy drop (0.04) in analog-aware training, compared to significant drops (up to 0.15) in pyramidal structures. Additionally, the paper emphasizes IMC's effective data pipelining, reducing latency and increasing throughput as well as the exploitation of inherent noise within AIMC, strategically harnessed to augment model certainty.
Skip Connections in Spiking Neural Networks: An Analysis of Their Effect on Network Training

H. Benmeziane A.Z. Ounnoughene I. Hamzaoui Y. Bouhadjar

ScaDL IPDPS 2023

Abs PDF

Spiking neural networks (SNNs) have gained attention as a promising alternative to traditional artificial neural networks (ANNs) due to their potential for energy efficiency and their ability to model spiking behavior in biological systems. However, the training of SNNs is still a challenging problem, and new techniques are needed to improve their performance. In this paper, we study the impact of skip connections on SNNs and propose a hyperparameter optimization technique that adapts models from ANN to SNN. We demonstrate that optimizing the position, type, and number of skip connections can significantly improve the accuracy and efficiency of SNNs by enabling faster convergence and increasing information flow through the network. Our results show an average +8% accuracy increase on CIFAR-10-DVS and DVS128 Gesture datasets adaptation of multiple state-of-the-art models. .