Number of hours
- Lectures 6.0
- Projects 0
- Tutorials 0
- Internship 0
- Laboratory works 0
- Written tests 0
ECTS
ECTS 0.0
Goal(s)
This course is an introduction to deep learning. The objective is both to get a global vision of this field in a short time and to introduce the concepts and tools that will be implemented in the Deep Learning projects (speech and audio project and image and hardware acceleration project).
Contact Laurent GIRINContent(s)
- The first part (2h) is an introduction to deep neural networks and machine learning. The course will first details machine learning in the case of linear models to reach the case of multilayer perception. The basics of optimization techniques such as gradient descent will be described. Some of the concepts in this course will be further discussed in the machine learning course of SICOM.
- The second part (2h) focuses on usual deep neural network models (FF-DNNs, CNNs, RNNs, CRNNs, etc.).
- The third part (2h) illustrates the use of deep learning in acoustic signal processing (speech and audio) and image processing, and will serve as an introduction to the speech and audio project and the image and hardware acceleration project that will use these techniques.
Prerequisites
- Basic knowledge of probability and statistics
- Basic knowledge of signal processing and information processing
Test
Only the projects will be evaluated
Cet enseignement n'est pas noté. La note est celle des Projets Audio ou Projets accélération matérielle qui accompagnent le cours.
Additional Information
Course list
Curriculum->Master->Semester 9
Curriculum->Double-Diploma Engineer/Master->Semester 9
Bibliography
Christopher Bishop, Pattern Recognition and Machine Learning. Springer, 2006.
Yann LeCun, Yoshua Bengio and Geoffrey Hinton, Deep learning. Nature, 521(7553), 436, 2015.