Architecture of neural networks for deep learning

Credits: 6

Semester: 3

Course: Core

Language of the course: English

Objectives

Students will study software development methodologies and programming technologies, methods and tools for developing and analyzing functional requirements for a project, theoretical foundations for in-depth training, modern technological solutions in the field of neural networks and in-depth training, modern tools for developing neural networks, a plan for developing system requirements , methods for evaluating and analyzing project results, methods and tools for organizing project data, methods for organizing neural networks, ways of us Royko neural networks, methods of preparation of test data sets, the components of the architecture of the neural network, regulatory documents, methods and tools for risk management, basic types of diagnostic data and methods of presentation.
Students will learn to apply methods and tools for analyzing functional requirements for an integration solution, apply methods and tools for developing technical specifications for an integration solution, make a choice of tools for developing neural networks; to select types of neural networks regarding project tasks, to make reports in accordance with existing standards, to apply software development methodologies, to apply software development project management methodologies, to use tools for joint development of neural networks, to interpret diagnostic data of solution operability, to apply risk management methods and tools .
Having finished the course, students will be able to: plan design work, select methodologies and templates, design neural network architecture, evaluate the state of analytical work as planned, describe the state of analytical work in a report format, select development tools, determine the set of libraries of reusable modules, deploy and configure collaborative development tools neural networks, assessing the quality of test datasets in accordance with the selected program and test method, are determined I applications of risk management, analysis and assessment of the risks, a choice of ways to respond to them and make available the necessary resources.

Contents

The main topics of the course are:

  • Artificial Neural Networks
  • Convolutional neural networks
  • Recurrent Neural Networks
  • Self-organizing cards
  • Boltzmann machines
  • Autocoders

Format

lectures and practical classes

Assessment

Examination