Enes Koşar - Academic Webpage

Research Interests

I have extensive experience in deep learning, specializing in novel ensemble approaches for time series signal classification for wearable sensor-based human activity and user identity recognition tasks. My work includes a thesis and journal papers in this area. Additionally, I have contributed to deep learning projects in the domains of computer vision and natural language processing.

Ph.D. Student, Bilkent University, Electrical and Electronics Engineering (2022 - Present)
M.Sc., Bilkent University, Electrical and Electronics Engineering (2019 - 2022)
B.Sc., Koç University, Computer Engineering (2017 - 2020)
B.Sc., Koç University, Electrical and Electronics Engineering (2014 - 2019)
Tokat Science High School (2009 - 2013)

Journal Publications
E. Koşar and B. Barshan, “A new cnn-lstm architecture for activity recognition employing wearable motion sensor data: Enabling diverse feature extraction,” Engineering Applications of Artificial Intelligence, vol. 124, p. 106529, 2023.
B. Barshan and E. Koşar, “Bidirectional Transfer Learning Between Activity and User Identity Recognition Tasks with a 2D CNN-LSTM Model for Wearables,” (under review in IEEE Internet of Things Journal).

Reviewed for Journal: Engineering Applications of Artificial Intelligence, Elsevier.

Context-aware object detection in aerial imagery with image inpainting
Human Activity recognition in videos using traditional computer vision methods
Face-mask detection by fine-tuning YOLOv5
Image captioning project with Python Keras using LSTM and pre-trained VGG16 networks
2D CNN network to score facial attractiveness of people with Python TensorFlow
Human Activity Recognition from time series signals using CNN, LSTM, and CNN-LSTM with Keras
Implementation of CNN, LSTM, GRU, RNN, ANN from scratch with Python
N-gram Language Model and basic NLP Implementations with Python