Many Computer Vision Lectures (full PDFs) are available below for free!
DESCRIPTION
Nowadays, Artificial Intelligence drives scientific and economic growth worldwide. This is largely due to advances in Machine Learning (ML), notably in Deep Neural Networks (DNNs), which are essentially massive ‘learning by experience/examples’ systems. Their applications span and revolutionize almost every human activity:
- Autonomous Systems (cars, drones, vessels),
- Media Content and Art Creation (including fake data creation/detection), Social Media Analytics,
- Medical Imaging and Diagnosis,
- Financial Engineering (forecasting and analytics), Big Data Analytics,
- Broadcasting, Internet and Communications,
- Robotics/Control
- Intelligent Human-Machine Interaction, Anthropocentric (human-centered)Computing,
- Smart Cities/Buildings and Assisted living.
- Scientific Modeling and Analytics.
Several DNN advances and challenges hit the news almost every day, arising discussions on AI ethics, privacy protection and its societal impact.
This CVML Web Module focuses on focuses on Machine Learning and Deep Neural Network theory, their applications in the above-mentioned diverse domains and new challenges ahead. As there is much hype and often little accuracy, when treating DNN topics, a rigorous mathematical treatment of all DNN topic is included in each lecture, focusing on both classification and regression problems.
The cornerstone DNN theories and technologies are presented: a) Artificial Neural Networks, Perceptron; b) Multilayer perceptron, Backpropagation; c) Convolutional Neural Networks (CNNs), both data classification and regression problems; d) Autoencoders; e) Recurrent Neural Networks. Applications follow in several image analysis, computer vision and autonomous system applications, notably: a) Deep learning for object detection, including special topics, e.g., on small object detection; b) Few-Shot Object Recognition; c) Deep Semantic Image Segmentation. Generative Adversarial Networks are presented that promise to revolutionize the way we create media/arts, while seriously threatening our democracy with fake data creation and spread. Deep Reinforcement Learning is also presented, as it is an essential element in novel Robotics/Control and other decision-making application domains.
LECTURE LIST
- Artificial Neural Networks. Perceptron
- Multilayer Perceptron. Backpropagation.
- Convolutional Neural Networks
- 1D Convolutional Neural Networks
- Deep Autoencoders
- Attention and Transformers Networks
- Recurrent Neural Networks. LSTMs
- Deep Object Detection
- Special topics in Object Detection
- Few Shot Object Recognition
- Deep Semantic Image Segmentation
- Adversarial Machine Learning
- Generative Adversarial Networks in Multimedia Creation
- Mathematical brain modeling
CVML WEB LECTURE MODULE SCHEDULE
This module has been designed to be mastered within 1 month (or less), if you have proper background (at least early undergraduate student in an EE, ECE, CS, CSE or any Engineering or Exact Sciences Department).
We propose that you follow the above mentioned Lecture order. You may want to skip few Lectures that might not be of immediate interest to you for later study.
On average you can study 4 lectures per week. The related effort is as follows:
1) Lecture pdf study and filling the related understanding questionnaire: 1-2 hours per lecture (on average, depending on your background)
2) Tutorial exercise (if available): 1/2 hour on average (more if you do not have theoretical skills). We strongly recommend to try solve them yourself, before resorting to the existing solution.
3) Programming exercise (if available): 3-4 hours on average (more if you do not have good programming skills). We strongly recommend to try program them yourself, before resorting to the existing code.
The following lectures are accompanied by programming or tutorial exercises:
- Convolutional Neural Networks (1 Programming exercise)
- Deep Object Detection (1 Programming exercise)