Demir, Alper

Profile Picture
Profile URL
https://hdl.handle.net/20.500.14365/6724
Name Variants
Demir, Alper
Job Title
Email Address
alper.demir@ieu.edu.tr
Main Affiliation
05.05. Computer Engineering
Status
Current Staff
Website
ORCID ID
0000-0003-2646-4850
Scopus Author ID
57549355800
Turkish CoHE Profile ID
717B71FC71E69FDA
Google Scholar ID
AJHWiu4AAAAJ
WoS Researcher ID
AAL-4037-2021
Files
Alper_Demir.jpg (9.51 KB)

Sustainable Development Goals

11

SUSTAINABLE CITIES AND COMMUNITIES
SUSTAINABLE CITIES AND COMMUNITIES Logo

1

Research Products

4

QUALITY EDUCATION
QUALITY EDUCATION Logo

0

Research Products

8

DECENT WORK AND ECONOMIC GROWTH
DECENT WORK AND ECONOMIC GROWTH Logo

0

Research Products

12

RESPONSIBLE CONSUMPTION AND PRODUCTION
RESPONSIBLE CONSUMPTION AND PRODUCTION Logo

0

Research Products

9

INDUSTRY, INNOVATION AND INFRASTRUCTURE
INDUSTRY, INNOVATION AND INFRASTRUCTURE Logo

1

Research Products

15

LIFE ON LAND
LIFE ON LAND Logo

0

Research Products

6

CLEAN WATER AND SANITATION
CLEAN WATER AND SANITATION Logo

0

Research Products

1

NO POVERTY
NO POVERTY Logo

0

Research Products

7

AFFORDABLE AND CLEAN ENERGY
AFFORDABLE AND CLEAN ENERGY Logo

0

Research Products

10

REDUCED INEQUALITIES
REDUCED INEQUALITIES Logo

0

Research Products

14

LIFE BELOW WATER
LIFE BELOW WATER Logo

0

Research Products

2

ZERO HUNGER
ZERO HUNGER Logo

0

Research Products

13

CLIMATE ACTION
CLIMATE ACTION Logo

0

Research Products

5

GENDER EQUALITY
GENDER EQUALITY Logo

0

Research Products

16

PEACE, JUSTICE AND STRONG INSTITUTIONS
PEACE, JUSTICE AND STRONG INSTITUTIONS Logo

0

Research Products

17

PARTNERSHIPS FOR THE GOALS
PARTNERSHIPS FOR THE GOALS Logo

1

Research Products

3

GOOD HEALTH AND WELL-BEING
GOOD HEALTH AND WELL-BEING Logo

0

Research Products
Documents

13

Citations

35

h-index

4

Go to Scopus profile
Documents

8

Citations

38

Go to WoS profile
Scholarly Output

8

Articles

2

Views / Downloads

7/18

Supervised MSc Theses

0

Supervised PhD Theses

0

WoS Citation Count

8

Scopus Citation Count

9

WoS h-index

2

Scopus h-index

2

Patents

0

Projects

5

WoS Citations per Publication

1.00

Scopus Citations per Publication

1.13

Open Access Source

3

Supervised Theses

0

Google Analytics Visitor Traffic

JournalCount
2024 Innovations in Intelligent Systems and Applications Conference, ASYU 2024 -- 2024 Innovations in Intelligent Systems and Applications Conference, ASYU 2024 -- 16 October 2024 through 18 October 2024 -- Ankara -- 2045622
19th International Conference on Compatibility Power Electronics and Power Engineering-CPE-POWERENG-Annual -- MAY 20-22, 2025 -- Antalya, TURKIYE1
3rd International Informatics and Software Engineering Conference, IISEC 20221
Applıed Intellıgence1
International Journal of Machine Learning and Cybernetics1
Current Page: 1 / 2

Scopus Quartile Distribution

Competency Cloud

GCRIS Competency Cloud

Filters

2022 - 20258
Reset filters

Settings

Scholarly Output Search Results

Now showing 1 - 8 of 8
  • Thumbnail Image
    Conference Object
    A Reinforcement Learning Based Approach to Solve Voltage Issues in Distribution Networks
    (IEEE, 2025) Cakir, Muhammed Turhan; Nayir, Hasan; Demir, Alper; Kaya, Huseyin; Ceylan, Oguzhan
    This paper proposes a Proximal Policy Optimization (PPO)-based reinforcement learning approach to solve over-voltage problem in power distribution networks. The approach aims to minimize the voltage deviations and to keep voltage magnitudes in the allowed ranges. The numerical simulations are performed on a modified unbalanced 123 node network. The modified test system includes a total number of 34 single phase Photovoltaics (200 kVA) connected to three phases. We modified the base case load profile based on real-world daily variations obtained from EPIAS. The PV generation profile was modeled according to a typical sunny day. Using OpenDSS and Python, we implemented PPO-based RL to optimize the setpoints of smart inverters and voltage regulators. The model was trained with load and solar profiles at 09:00, 12:00, and 16:00 to derive optimal voltage regulation strategies for these time points. From the simulation results, we observed that the proposed PPO-based RL approach significantly reduces voltage deviations across all phases, which may help efficient operation of the distribution networks.
  • Thumbnail Image
    Research Project
    Kısmi Gözlemlenebilir Pekiştirmeli Öğrenmede Faydalı Bellek Oluşturma
    (2023) Demirbilek, Burak Han; Demir, Alper
    Pekiştirmeli öğrenme, gerçek hayattaki bir öğrenme ortamını modellemeyi amaçlayan önemli bir makine öğrenme tekniğidir. Konuyla ilgili son araştırmalar, araştırmacılar arasında onu çok popüler hale getirmiş ve birçok gerçek hayat senaryosuna daha uygulanabilir olmasını sağlamıştır. Kısmi gözlemlenebilirlik altında pekiştirmeli öğrenme, çevreden toplanan sınırlı bilgi nedeniyle özellikle zorlu bir alandır. Bu gibi hallerde, etmen, verilen görev için bir hareket tarzı oluşturabilmesi için ortamın durumunu tahmin etmek için deneyimlerden oluşan bir bellek tutmak zorundadır. Bazı çalışmalar, etmenin mevcut andan önceki bir dizi gözlemi ve eylemi bellekte tutmasını sağlayan basit bir pencere tabanlı bellek yaklaşımı kullanır, ancak bu tür yöntemler, bir bilgiyi uzun süre boyunca bellekte tutmayı gerektirdiği problemlere genellenemez. Diğer çalışmalar, gerekli bilgileri yapı içinde tutan karmaşık bir model kullanır, ancak bu tür modeller, probleme çok özel olmak ve analiz için çok kapalı olmaktan dolayı yetersiz kalmaktadırlar. Bu tür problemlerde, bir eyleme karar verecek ilgili bilgi zamansal olarak uzaktır, bu nedenle etmen neyi bellekte tutacağı konusunda seçici olmak zorundadır. Yararlı bir bellek tutma problemini ele alan çalışma azdır ve uygulanabilirlikleri sınırlıdır. Bu sebeple bu alan keşfedilmemiş kalmıştır. Bu çalışmada, belleği değiştiren eylemler yapmasına izin vererek bellek kontrolünü etmene verme fikrini takip ediyoruz. Böylece etmen, bir ortamın dinamiklerine daha uyumlu hale gelir. Ayrıca, bu öğrenme mekanizmasını desteklemek için, etmenin ayırt edici olayları hatırlamasında yol gösterici olan ve ortamdaki durumunu netleştirmesini sağlayan bir içsel motivasyon yapısı oluşturduk. Genel yaklaşımımız, birkaç pekiştirmeli öğrenme yöntemine uygulanmış, uzun süreli bellek gerektiren birkaç kısmi gözlemlenebilir problem üzerinde test edilmiş ve analiz edilmiştir. Deneyler, diğer bellek tabanlı yöntemlere kıyasla öğrenme performansı açısından net bir gelişme göstermektedir.
  • Thumbnail Image
    Article
    Citation - WoS: 2
    Citation - Scopus: 3
    Learning What To Memorize: Using Intrinsic Motivation To Form Useful Memory in Partially Observable Reinforcement Learning
    (Springer, 2023) Demir, Alper
    Reinforcement Learning faces an important challenge in partially observable environments with long-term dependencies. In order to learn in an ambiguous environment, an agent has to keep previous perceptions in a memory. Earlier memory-based approaches use a fixed method to determine what to keep in the memory, which limits them to certain problems. In this study, we follow the idea of giving the control of the memory to the agent by allowing it to take memory-changing actions. Thus, the agent becomes more adaptive to the dynamics of an environment. Further, we formalize an intrinsic motivation to support this learning mechanism, which guides the agent to memorize distinctive events and enable it to disambiguate its state in the environment. Our overall approach is tested and analyzed on several partial observable tasks with long-term dependencies. The experiments show a clear improvement in terms of learning performance compared to other memory based methods.
  • Thumbnail Image
    Conference Object
    Analyzing Traffic Patterns in Izmir: a Study on Busy Hours and Congestion
    (Institute of Electrical and Electronics Engineers Inc., 2024) Başoğul, Ali Ozan; Çekirdek, Hamza; Nakipoğlu, K.; Yağcı, Semih; Demir, Alper
    Developing strategies for transportation is one of the main tasks for smart cities. With the traffic data on key arteries of Izmir, this project aims to estimate the time intervals and routes where traffic jam may occur in Izmir and uncover relationships between various factors affecting traffic. The study demonstrates that accurate models can be developed to predict the number of passing vehicles and reveals interesting correlations within the data. © 2024 IEEE.
  • Thumbnail Image
    Correction
    Landmark Based Guidance for Reinforcement Learning Agents Under Partial Observability (nov 2022, 10.1007/S13042-022-01713-5)
    (Springer Heidelberg, 2023) Demir, Alper; Çilden, Erkin; Polat, Faruk
    [No Abstract Available]
  • Thumbnail Image
    Conference Object
    Water Consumption Dynamics in Izmir: Analyzing Influences of District, Seasonality, and External Events
    (Institute of Electrical and Electronics Engineers Inc., 2024) Arslan, A.; Titiz, I.E.; Gürcan, E.C.; Demir, A.
    This study investigates the patterns of water consumption in Izmir from January 2015 to January 2024, using a dataset provided by Izmir's Open Data Portal. The research aims to understand how different factors such as district, seasons, and external events like the COVID-19 pandemic influence water usage across the city. By employing statistical analysis and machine learning models including Decision Trees and K-means, this project identifies significant spatial and temporal variations in water consumption. This analysis not only aids in the efficient management of water resources but also serves as a foundation for future predictive modeling and sustainability efforts in urban settings. © 2024 IEEE.
  • Thumbnail Image
    Conference Object
    Behaviour Analysis of Izmir Residents Using Public Wi-Fi Access Point Usage
    (Institute of Electrical and Electronics Engineers Inc., 2022) Dilek I.; Oguz K.; Demir A.
    The wide distribution of access points in Izmir allows the collected information to be employed in smart city algorithms. In this study, we analyze the information that has been made publicly available by Izmir Metropolitan Municipality. We first show that the data is reliable, then analyze it from the perspectives of holidays, seasonal trends, and the COVID-19 pandemic. The study also shows that the information can be used for crowd analysis and forecasting, using K-means and SARIMA algorithms, respectively. © 2022 IEEE.
  • Thumbnail Image
    Article
    Citation - WoS: 6
    Citation - Scopus: 6
    Landmark Based Guidance for Reinforcement Learning Agents Under Partial Observability
    (Springer Heidelberg, 2023) Demir, Alper; Cilden, Erkin; Polat, Faruk
    Under partial observability, a reinforcement learning agent needs to estimate its true state by solely using its observation semantics. However, this interpretation has a drawback, which is called perceptual aliasing, avoiding the convergence guarantee of the learning algorithm. To overcome this issue, the state estimates are formed by the recent experiences of the agent, which can be formulated as a form of memory. Although the state estimates may still yield ambiguous action mappings due to aliasing, some estimates exist that naturally disambiguate the present situation of the agent in the domain. This paper introduces an algorithm that incorporates a guidance mechanism to accelerate reinforcement learning for partially observable problems with hidden states. The algorithm makes use of the landmarks of the problem, namely the distinctive and reliable experiences in the state estimates context within an ambiguous environment. The proposed algorithm constructs an abstract transition model by utilizing the landmarks observed, calculates their potentials throughout learning -as a mechanism borrowed from reward shaping-, and concurrently applies the potentials to provide guiding rewards for the agent. Additionally, we employ a known multiple instance learning method, diverse density, for automatically discovering landmarks before learning, and combine both algorithms to form a unified framework. The effectiveness of the algorithms is empirically shown via extensive experimentation. The results show that the proposed framework not only accelerates the underlying reinforcement learning methods, but also finds better policies for representative benchmark problems.