2024/2025
Modern Concepts of Dynamical Neural Networks
Type:
Mago-Lego
Delivered by:
School of Data Analysis and Artificial Intelligence
When:
1, 2 module
Open to:
students of all HSE University campuses
Language:
English
ECTS credits:
6
Contact hours:
54
Course Syllabus
Abstract
This course delves into the modern concepts of Dynamic Neural Networks (DNNs), representing the forefront of artificial intelligence and deep learning innovation. Dynamic Neural Networks stand out with their ability to adapt their architecture and functionality dynamically, responding to varying inputs and operational conditions. Students will explore state-of-the-art techniques such as dynamic routing, layer and
neuron adaptation, and advanced attention mechanisms, including transformers. The course also covers cutting-edge topics like meta-learning, few-shot learning, and neural architecture search (NAS), which enable DNNs to optimize their structure and learning processes in real-time. Additionally, the course will address the implementation of
resource-aware and sparse networks, crucial for deploying AI on evices with limited computational capabilities. Through a blend of theoretical knowledge and hands-on projects, participants will gain insights into how these modern DNN concepts are revolutionizing applications in computer vision, natural language processing, robotics, and ersonalized recommendations. This course is ideal for AI researchers, practitioners,
and enthusiasts eager to master the latest advancements in dynamic neural network technology.