Loading...

Reinforcement Learning Approach in Self-Assembly Systems to Acquire Desired Structures

Ravari, Amir Hossein | 2021

356 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53729 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Bagheri Shouraki, Saeed
  7. Abstract:
  8. Self-Assembly (SA) plays a critical role in the formation of different phenomena in nature. This phenomenon can be defined as an arrangement of meaningful patterns with the aggregate behavior of simpler structures. One of the examples of Self-Assembly can be considered of the formation of ice crystals from ice molecules. Previous works mainly focus on graph grammar and self-assembly in fully observable environments. These algorithms mainly consist of two main stages: first, constructing simpler structures and then joining these simpler structures to form a complex structure. The challenges of the previous work can be considered as the necessity of a central controller in the formation of Self-Assembly and also the complexity of the algorithms for constructing a complex structure. We tend to explore Reinforcement Learning (RL) in self-assembling systems in partially observable environments to form decentralized learning for the environment's agents. This article proposes a new Q-Learning method to correct the Q-values based on constant goal change in a partially observable environment. We also offer an original algorithm for balancing exploration and exploitation based on the idea of ART neural networks. Also, to solve the scalability issue, we propose the concept of super-agents that can help us build more complex structures using local ones. Our algorithm is capable of forming structures in a partially observable environment. It is also simpler to program each agent using this algorithm. The agents can learn from each other and measure the quality of their interaction with the environment to achieve the desired results. The results consist of implementing the previous research results with our algorithm and forming complex structures
  9. Keywords:
  10. Self Assembly ; Modeling ; Reinforcement Learning ; Multi-Agent Reinforcement Learning ; Q-Learning

 Digital Object List

 Bookmark

No TOC