Onsite Deep Neural Network Training
The goal of the course is to give a deep understanding of deep neural networks and their capabilities.
Environment set up: VS.Code, PIP, VirtualEnv, Jupyter Notebook, Google Colab
2.1 Using pre-trained networks and third party libraries:
2.1.1 Colorizing images using Colab
2.1.2 Style transfer using Vs.Code
2.1.3 Generating faces using Colab
2.1.4 Generating and summarizing texts using Colab or Vs.Code
3.1 Using Fast.ai and resNet50 with transferlearning
4.1 Our goal is to design a neural network to recognize images
4.2 How can we feed an image to a network?
4.3 What is a network after all? - vectors, matrix, linear algebra
4.4 Backpropagation, activation functions
4.5 We will implement our very first neural nets without any library
4.6 Numpy, Tensors
4.7 Recognizing handwritten digits -MNIST
5.1 Convolutional Neural Network using PyTorch
5.2 Creating image dataset from the web
5.3 Using datasets in PyTorch
5.4 Comparing pre-trained and from scratched networks by implementing both
6.1 One-hot encoding
6.3 Character level RNN+ LSTM
6.4 Generating text with RNN+LSTM
Questions for organizers
It can be either english or hungarian.
1 week, but It depends of the group. If we are too slow I will skip parts or implementation details, if we are too fast we will start with Reinforcement learning as well.
The date will be discussed with the customer.
At the customer site. Around Budapest, Hungary or Bonn, Germany is preferred. Anything else is up for discussion.
Contact us to discuss it.
Technical questions for the students
Python knowledge is necessary for the course. We can offer free online Python course if it is needed, currently in hungarian. Or in english/hungarian onsite, contact for more details.
It does not matter if you are choosing between Linux, Windows or MAC :)
Yes, bring your favorite laptop or desktop computer :)
This course page is just a placeholder for providing detailed information of the deep neural network onsite training. By subscribing to it means absolutely nothing. To order the onsite training please contact us.