Overview

I currently work on cryptography and machine learning, with a focus on efficient and privacy-preserving machine learning systems.

Main Research Interests

Efficient Machine Learning

I am mainly interested in efficient machine learning, more precisely on quantization and quantized neural networks and their applications. This includes:

• Quantization techniques: Developing methods to reduce the precision of neural network weights and activations while maintaining performance
• Hardware-specific optimizations: Adapting neural network architectures for efficient deployment on specialized hardware

Privacy-Preserving Machine Learning (PPML)

A significant part of my research focuses on privacy-preserving machine learning, particularly:

• Full Homomorphic Encryption (FHE): Designing neural network architectures that are compatible with FHE schemes, enabling computation on encrypted data
• Privacy in Machine Learning: Exploring methods to train and deploy machine learning models while protecting sensitive data

Explainable AI (XAI)

I am also interested in making machine learning models more interpretable and explainable:

• Interpretable neural network architectures: Developing models that are both accurate and interpretable
• Rule-based models: Creating neural network-based rule models that provide clear decision-making processes

Current Work

My PhD research, under the supervision of Thomas Peyrin, focuses on bridging the gap between cryptography and machine learning, with particular emphasis on:

• Designing FHE-friendly neural network architectures
• Developing scalable and verifiable neural networks
• Creating interpretable models for real-world applications

For more details about my publications, please see my Resume or visit my Homepage.