Our Research
Research work by Assem Sabry and the TokenAI team. Our focus spans foundational AI algorithms, deep learning optimizers, and applied medical imaging models.
2026
STAM Optimizer
Stable Training with Adaptive Momentum — a JAX-based optimizer that dynamically adapts the first-order momentum coefficient β₁ based on online gradient variance, achieving top-3 performance on 83% of benchmark phases.
Published: 1 May 2026
AI Comes at a Cost
The paper "AI Comes at a Cost" examines both the transformative benefits of Artificial Intelligence and its hidden cognitive, social, and economic risks. It argues that heavy reliance on AI may lead to reduced human abilities such as critical thinking, creativity, memory, and problem-solving, alongside growing dependency on AI systems in decision-making.
Published: 20 May 2026
2025
Histopathological AI Detection
HAID is a deep learning AI model that detects breast cancer in histopathology images, helping doctors and labs achieve faster and more accurate diagnoses.
Published: 22 June 2025
Intelligent Brain Tumor Detector
IBTD is a deep learning model that identifies and classifies brain tumors from MRI scans using a fine-tuned ResNet50 architecture and advanced training methods for higher accuracy.
Published: 14 July 2025
2024
No publications available for this year.