EEE581 Special Topics in Deep Learning for Signal and Image Analysis
| Course Code: | 3560581 |
| METU Credit (Theoretical-Laboratory hours/week): | 3 (3.00 - 0.00) |
| ECTS Credit: | 8.0 |
| Department: | Electrical and Electronics Engineering |
| Language of Instruction: | English |
| Level of Study: | Masters |
| Course Coordinator: | |
| Offered Semester: | Fall and Spring Semesters. |
Course Objectives
Course Content
This course provides an exploration of deep learning techniques for analysing signals and images across diverse domains. Key topics include signal representations in time and frequency domains, neural networks, and advanced deep learning models such as RNNS, CNNs, Autoencoders, and Transformers. Applications range from audio classification, biomedical signal processing, and time-series prediction to computer vision. Class sessions will include both theoretical instruction and practical examples. Students will gain hands-on experience through coding assignments using TensorFlow/Keras, preparing them for advanced research and industry challenges in Al-driven signal and image analysis.
The course is organised to be offered to students from diverse engineering fields. The course can be taken by students in Electrical and Electronics Engineering, Computer Engineering, SEES, and Mechanical Engineering programs.
- This course introduces different signal and image representation techniques, such as time-frequency representations, spectrograms, and feature extraction methods, that can be input to deep learning models.
- This course presents the major concepts Neural Networks (NNs) including biological neuron, artificial neuron (perceptron), neural network topologies and learning in NNs.
- Then, we will study the latest deep neural network architectures for different tasks such as classification, detection and prediction with a special focus on audio signals, images, and videos.
- In particular, we will focus on both theory and practice of Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs) - particularly Long-Short-Term Memory Networks (LSTM), Autoencoders and Transformers.
- Students will have both theoretical and practical knowledge with hands-on experience through coding the assignments using TensorFlow/Keras.