Specialization: AI / Computer Vision
Paper Title: An Efficient Human Face Expression Recognition Based on Deep Learning Algorithm
Brief Description:
This study proposes a hybrid facial expression recognition system combining machine learning and deep learning. It detects and focuses on key facial regions (eyes and mouth) using traditional methods, then applies a customized CNN with multiple color-space inputs to classify seven expressions. Tested on KDEF, JAFFE, and FER2013 datasets, the system achieved over 99% accuracy, outperforming AlexNet and VGG16.
Paper Title: Continuous Learning for Automated Early-Stage Alzheimer’s Detection Using MRI and Spike Neural Signals
Brief Description:
This study presents a continual learning framework that combines MRI and EEG data for early Alzheimer’s disease (AD) detection. Using multimodal deep learning with mechanisms to prevent catastrophic forgetting, the system continuously learns from new patient data while retaining previous knowledge. Among the fusion strategies tested, mid-fusion achieved the best performance, reaching 86% accuracy, demonstrating the effectiveness of combining structural MRI and temporal EEG information for early AD diagnosis.
Paper Title: A Customized CNN for Multi-Class Autism Classification Using the CARS-ADK Dataset
Brief Description:
This study developed a new ASD dataset (CARS-ADK) in the Kurdistan Region and evaluated several machine learning and deep learning models for three-class ASD classification. Results showed that all models performed well, with the customized CNN achieving the highest performance, demonstrating the potential of AI-assisted ASD diagnosis.
Presented by: Dr. Salar Jamal Abdulhameed on 17th of June 2026
