AI HOUSE and Lviv Polytechnic launch a new offline school Deep Neutral Networks for Computer Vision Tasks

Department of Artificial Intelligence Systems, Lviv Polytechnic
Заставка школи

AI HOUSE and Lviv Polytechnic National University are launching a new offline school Deep Neural Networks for Computer Vision Tasks. This is an offline school on the basic approaches to learning neural networks, using convolutional neural network models, configure shallow neural networks, Python, OpenCV, Keras, PyTorch, Scikit-image and Jupyter Notebook. Classes will take place in the Lviv Polytechnic Student Library.

The school is organized by AI HOUSE – a non-profit organization whose main goal is to build the largest and most powerful AI community in Ukraine for the growth of new AI startups, and the development of the field of artificial intelligence and the technology sector in general. The project is part of the ecosystem of the technology company Roosh and unites AI/ML specialists in its initiatives, such as machine learning schools, workshops and meetups with practicing lecturers, conferences with international specialists, AI breakfasts to exchange ideas with AI thought leaders, etc.

In February, the team held a one-day workshop Computer Vision: Roof Footprints Recognition in the field of computer vision. This time the training will take place in Lviv and will last four weeks.

Learn more about the school

The school will be useful for specialists who have more than one year of experience in machine learning, Python and have a basic knowledge of PyTorch and Keras. Classes at the school will take place offline in Lviv every Saturday in the morning and will last for four weeks, from April 22 to May 20, 2023. At the end of the school, students will learn to detect emotions of dogs in photos using neural networks.

The teacher is Anastasia Deineko, Technical Trainer at Grid Dynamics, Candidate of Technical Sciences, Associate Professor at IT STEP University in Lviv.

What will you learn?

During their study students will learn how to use pre-trained convolutional network models, learn about overtraining and undertraining problems, will understand the basics of fine-tuning convolutional networks, and improve their programming skills.

The school program involves the following topics:

  1. Introduction to neural networks – from shallow to deep;
  2. Learning paradigms – learning with a teacher and self-learning;
  3. Fundamentals of optimization theory (gradient optimizers);
  4. Multilayer perceptron, activation functions, vanishing and exploding gradient problem;
  5. Computer vision:
    • Neural networks, architectures of convolutional networks, transfer learning, image classification;
    • Autocoders, VAE, CVAE;
    • Detection of objects;
    • Data segmentation;
    • Generative adversarial networks (GAN);
    • Meta-learning: metric learning, domain adaptation;
    • Weak learning, active learning,
  6. Recurrent neural networks (LSTM, GRU);
  7. From sequence to sequence models (Transformers).

How to participate?

To become a student of Deep Neural Networks for Computer Vision Tasks, you need to register.

Participation in the school will be provided for any donation from 400 hryvnias. The collected funds will be transferred to the reconstruction of the building of the Faculty of Computer Science and Cybernetics, Taras Shevchenko National University of Kyiv.

The number of places is limited – up to 25 students can join the school. Priority will be given to more motivated students with relevant skills and background.

Registration for the course will last until April 16.