Deep Learning for Bioinformatics

This course addresses the concepts and hands-on implementation of neural networks as supervised techniques for analysis, classification, and prediction customized for bioinformatics and medical image processing applications.

Course code: DL-BIO

Instructor: Prof. Dr. Mohammed A.-Megeed Salem

Professor of Image & Vision Computing, German University in Cairo, GUC

Workload: 12 lectures, 3 hours each. Total workload: 42 hours: 36 hours of lectures and tutorials and 6 hours of self studies.

Entrance requirements: Computer science. Basic knowledge of Machine Learning.

Used media: PowerPoint presentation

Objectives

• Understand neural networks and deep learning applications in bioinformatics and medical image processing
• Apply deep learning to analysis, classification, prediction, medical image classification, and segmentation

Competences to be Developed

• Simple supervised machine learning using conventional approaches
• Binary classification of biomedical data using a single neuron
• Feedforward neural networks for multi-class classification
• Backpropagation algorithms
• Hyperparameter tuning and neural network optimization
• Building simple deep learning models for medical image classification
• Adopting pretrained architectures for medical image segmentation and genomic data clustering

Assessment

• Finalize a research project applying learned methods
• Compile project outcomes into a high-quality research article ready for peer-review submission
• Present and scientifically review all projects in the last lecture

Assessment and completion

• Quizzes: 20%
• Assessments: 30%
• Final / Graduation Project: 50%
• Overall passing grade: 60%
• Minimum project grade: 50%
• All mandatory components must be completed
• One reassessment or resubmission opportunity is available for a failed mandatory assessment or final project