Neural Networks and Deep Learning
***
The first lecture is scheduled on January 11, 2022 at 9:00.
All lectures will be in presence in the GRAY ROOM of the TeCIP Institute, on each Monday and Tuesday, from 9:00 to 11:30.
Registration
To attend the lectures and receive notifications about any change on course lectures,
please register in this file.
This course (6 CFUs) includes two modules of 3 CFUs each: the first module focuses on the theoretical foundations
of neural networks and deep learning, while the second module covers more practical and implementation issues.
Course Program
Course topics: theoretical part
- Introduction to neural computing
- Hopfield networks
- Unsupervised learning
- Reinforcement learning
- Supervised learning - part 1
- Supervised learning - part 2
- Towards Deep Neural Networks (DNNs)
- Autoencoders
- Convolutional Neural Networks
- DNNs for object classification
- DNNs for object detection
- Recurrent Neural Networks
- Deep Reinforcement Learning
- Generative Adversarial Networks
- Sample applications and open issues
Slides on the practical part
- Implementing Hopfield networks in C
- Implementing Kohonen networks in C
- Implementing Reinforcement Learning in C
- Implementing Backpropagation in C
- Frameworks for training and inference of DNNs
- Modeling DNNs in Tensorflow and Caffe
- Overview of the Apollo framework for Autonomous Driving
- Networks in Apollo: Perception Module
- Neural-based control: OpenAI Gym simulation environments
- Deep Reinforcement Learning: DDPG
- Genetic Algorithms for control by Reinforcement Learning
- Adversarial examples: attacks and defense techniques
- Implementing DNNs on GPUs platforms - Part I
- Implementing DNNs on GPUs platforms - Part II
- Accelerating DNNs on FPGA platforms
- The CHaiDNN framework
- The Vitis AI framework
Projects