Sürdürülebilir Yeşil Kampüs Koleksiyonu / Sustainable Green Campus Collection

Permanent URI for this collectionhttps://hdl.handle.net/11147/7755

Browse

Filters

2003 - 200912010 - 20171

Settings

Access Right: Open AccessAuthor: search.filters.author.Özdemir, Serhan

Search Results

Now showing 1 - 2 of 2
  • Article
    Citation - WoS: 6
    Citation - Scopus: 7
    An Entropy-Based Analysis of Lane Changing Behavior: An Interactive Approach
    (Taylor and Francis Ltd., 2017) Koşun, Çağlar; Özdemir, Serhan; Özdemir, Serhan; 03.10. Department of Mechanical Engineering; 03. Faculty of Engineering; 01. Izmir Institute of Technology
    Objectives: As a novelty, this article proposes the nonadditive entropy framework for the description of driver behaviors during lane changing. The authors also state that this entropy framework governs the lane changing behavior in traffic flow in accordance with the long-range vehicular interactions and traffic safety. Methods: The nonadditive entropy framework is the new generalized theory of thermostatistical mechanics. Vehicular interactions during lane changing are considered within this framework. The interactive approach for the lane changing behavior of the drivers is presented in the traffic flow scenarios presented in the article. According to the traffic flow scenarios, 4 categories of traffic flow and driver behaviors are obtained. Through the scenarios, comparative analyses of nonadditive and additive entropy domains are also provided. Results: Two quadrants of the categories belong to the nonadditive entropy; the rest are involved in the additive entropy domain. Driving behaviors are extracted and the scenarios depict that nonadditivity matches safe driving well, whereas additivity corresponds to unsafe driving. Furthermore, the cooperative traffic system is considered in nonadditivity where the long-range interactions are present. However, the uncooperative traffic system falls into the additivity domain. The analyses also state that there would be possible traffic flow transitions among the quadrants. This article shows that lane changing behavior could be generalized as nonadditive, with additivity as a special case, based on the given traffic conditions. Conclusions: The nearest and close neighbor models are well within the conventional additive entropy framework. In this article, both the long-range vehicular interactions and safe driving behavior in traffic are handled in the nonadditive entropy domain. It is also inferred that the Tsallis entropy region would correspond to mandatory lane changing behavior, whereas additive and either the extensive or nonextensive entropy region would match discretionary lane changing behavior. This article states that driver behaviors would be in the nonadditive entropy domain to provide a safe traffic stream and hence with vehicle accident prevention in mind.
  • Master Thesis
    Data Driven Modeling Using Reinforcement Learning in Autonomous Agents
    (Izmir Institute of Technology, 2003) Karakurt, Murat; Özdemir, Serhan; Özdemir, Serhan; 03.10. Department of Mechanical Engineering; 03. Faculty of Engineering; 01. Izmir Institute of Technology
    This research has aspired to build a system which is capable of solving problems by means of its past experience, especially an autonomous agent that can learn from trial and error sequences. To achieve this, connectionist neural network architectures are combined with the reinforcement learning methods. And the credit assignment problem in multi layer perceptron (MLP) architectures is altered. In classical credit assignment problems, actual output of the system and the previously known data in which the system tries to approximate are compared and the discrepancy between them is attempted to be minimized. However, temporal difference credit assignment depends on the temporary successive outputs. By this new method, it is more feasible to find the relation between each event rather than their consequences.Also in this thesis k-means algorithm is modified. Moreover MLP architectures is written in C++ environment, like Backpropagation, Radial Basis Function Networks, Radial Basis Function Link Net, Self-organized neural network, k-means algorithm.And with their combination for the Reinforcement learning, temporal difference learning, and Q-learning architectures were realized, all these algorithms are simulated, and these simulations are created in C++ environment.As a result, reinforcement learning methods used have two main disadvantages during the process of creating autonomous agent. Firstly its training time is too long, and too many input parameters are needed to train the system. Hence it is seen that hardware implementation is not feasible yet. Further research is considered necessary.