• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
For visually-impairedUser profile (HSE staff only)SearchMenu

Methods of Formal Concepts Analysis as a Tool for Designing the Architecture of Neural Networks for Non-Player Characters in Unity

Student: Sergei Ermakov

Supervisor: Sergei Kuznetsov

Faculty: Faculty of Computer Science

Educational Programme: Statistical Learning Theory (Master)

Final Grade: 7

Year of Graduation: 2021

To investigate the possibilities of applying formal concept analysis to deep reinforcement learning, we took the task of teaching a non player character (NPC) certain actions that we want. The training was based on a neural network model. We applied the formal concepts analysis (FCA) to its construction. In order to investigate the advantages of such a model, we carried out a number of experiments in a training environment, with various models of neural networks (4 different architectures), which, as a result of training, as an output value produced an action leading to the greatest reward in the current state. The first and second chapters of this work present the theoretical foundations of reinforcement learning, the analysis of formal concepts, and describe the main components of the learning environment involved in the reinforcement learning process on the Unity platform. The third chapter presents the results of training four different models of neural networks, performing the task in a learning environment. A comparison is made with the results obtained using a neural network based on the method of formal concepts analysis. The result of our work showed that a neural network based on the analysis of formal concepts successfully solves the task and, with an increase in the time for each training episode, the number of episodes in general required to achieve the training goal becomes minimal, in comparison with three other models of neural networks.

Full text (added May 27, 2021)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses