Koç University Computer Science MSc 2020-2022
Koç University Computer Science PhD 2022-2025
RESEARCH, TEACHING, or OTHER INTERESTS
Computer Vision and Pattern Recognition
7
Scopus Publications
343
Scholar Citations
8
Scholar h-index
5
Scholar i10-index
Scopus Publications
SLOT-GUIDED ADAPTATION OF PRE-TRAINED DIFFUSION MODELS FOR OBJECT-CENTRIC LEARNING AND COMPOSITIONAL GENERATION 13th International Conference on Learning Representations Iclr 2025, 2025
ADAPT: Efficient Multi-Agent Trajectory Prediction with Adaptation Görkay Aydemir, Adil Kaan Akan, Fatma Güney Proceedings of the IEEE International Conference on Computer Vision, 2023 Forecasting future trajectories of agents in complex traffic scenes requires reliable and efficient predictions for all agents in the scene. However, existing methods for trajectory prediction are either inefficient or sacrifice accuracy. To address this challenge, we propose ADAPT, a novel approach for jointly predicting the trajectories of all agents in the scene with dynamic weight learning. Our approach outperforms state-of-the-art methods in both single-agent and multi-agent settings on the Argoverse and Interaction datasets, with a fraction of their computational overhead. We attribute the improvement in our performance: first, to the adaptive head augmenting the model capacity without increasing the model size; second, to our design choices in the endpoint-conditioned prediction, reinforced by gradient stopping. Our analyses show that ADAPT can focus on each agent with adaptive prediction, allowing for accurate predictions efficiently. https://KUIS-AI.github.io/adapt
SLAMP: Stochastic Latent Appearance and Motion Prediction Adil Kaan Akan, Erkut Erdem, Aykut Erdem, Fatma Guney Proceedings of the IEEE International Conference on Computer Vision, 2021 Motion is an important cue for video prediction and often utilized by separating video content into static and dynamic components. Most of the previous work utilizing motion is deterministic but there are stochastic methods that can model the inherent uncertainty of the future. Existing stochastic models either do not reason about motion explicitly or make limiting assumptions about the static part. In this paper, we reason about appearance and motion in the video stochastically by predicting the future based on the motion history. Explicit reasoning about motion without history already reaches the performance of current stochastic models. The motion history further improves the results by allowing to predict consistent dynamics several frames into the future. Our model performs comparably to the state-of-the-art models on the generic video prediction datasets, however, significantly outperforms them on two challenging real-world autonomous driving datasets with complex motion and dynamic background.
Just Noticeable Difference for Machines to Generate Adversarial Images Adil Kaan Akan, Mehmet Ali Genc, Fatos T. Yarman Vural Proceedings International Conference on Image Processing Icip, 2020 One way of designing a robust machine learning algorithm is to generate authentic adversarial images that can trick the algorithms as much as possible. In this study, we propose a new method to generate adversarial images which are very similar to true images, yet, these images are discriminated from the original ones and are assigned into another category by the model. The proposed method is based on a popular concept of experimental psychology, called, Just Noticeable Difference. We define Just Noticeable Difference for a machine learning model and generate the least perceptible difference for adversarial images which can trick a model. The suggested model iteratively distorts a true image by gradient descent method until the machine learning algorithm outputs a false label. Deep Neural Networks are trained for object detection and classification tasks. The cost function includes regularization terms to generate just noticeably different adversarial images which can be detected by the model. The adversarial images generated in this study look more natural compared to the output of the state of the art adversarial image generators.
Modeling and decoding complex problem solving process by artificial neural networks Adil Kaan Akan, Baran Baris Kivilcim, Emre Akbas, Sharlene D. Newman, Fatos T. Yarman Vural 27th Signal Processing and Communications Applications Conference Siu 2019, 2019 It is hypothesized that the process of complex problem solving in human brain consists of two basic phases, namely, planning and execution. In this study, we propose a computational model in order to verify this hypothesis. For this purpose, we develop a holistic approach for decoding the planning and execution phases of complex problem solving, using the functional magnetic resonance imaging data (fMRI), recorded when the subjects play the Tower of London (TOL) game. In the first step of the proposed study, we estimate a brain network, called Artificial Brain Network (ABN), by designing an artificial neural network, whose weights correspond to the edge weights of the brain network established among the anatomic regions. Then, we decode the planning and execution tasks of complex problem slowing by training a multi-layer perceptron. It is shown that the edge weights of the artificial brain network capture the functional connectivity among anatomic brain regions. When trained on the edge weights of brain networks extracted from average BOLD activation of anatomical regions, the proposed model successfully discriminates the planning and execution phases of complex problem solving process. We compare the suggested computational brain network model to the state of the art models reported in the literature and observe that the decoding performance of the suggested model is better then the available methods in the literature.
RECENT SCHOLAR PUBLICATIONS
Aligning Latent Geometry for Spherical Flow Matching in Image Generation THS Meral, K Oktay, H Yesiltepe, AK Akan, P Yanardag arXiv preprint arXiv:2605.15193 , 2026 2026
Infinity-rope: Action-controllable infinite video generation emerges from autoregressive self-rollout H Yesiltepe, THS Meral, AK Akan, K Oktay, P Yanardag arXiv preprint arXiv:2511.20649 , 2025 2025 Citations: 24
Learning Object-Centric Representations Based on Slots in Real World Scenarios AK Akan arXiv preprint arXiv:2509.24652 , 2025 2025
Compositional video synthesis by temporal object-centric learning AK Akan, Y Yemez arXiv preprint arXiv:2507.20855 , 2025 2025 Citations: 2
Slot-guided adaptation of pre-trained diffusion models for object-centric learning and compositional generation Y Yemez International Conference on Learning Representations 2025, 93113-93126 , 2025 2025 Citations: 9
Adapt: Efficient multi-agent trajectory prediction with adaptation G Aydemir, AK Akan, F Güney Proceedings of the IEEE/CVF International Conference on Computer Vision … , 2023 2023 Citations: 135
Stretchbev: Stretching future instance prediction spatially and temporally AK Akan, F Güney European Conference on Computer Vision, 444-460 , 2022 2022 Citations: 72
Stochastic future prediction in real world driving scenarios AK Akan arXiv preprint arXiv:2209.10693 , 2022 2022 Citations: 1
Just noticeable difference for machine perception and generation of regularized adversarial images with minimal perturbation AK Akan, E Akbas, FTY Vural Signal, Image and Video Processing 16 (6), 1595-1606 , 2022 2022 Citations: 6
Trajectory forecasting on temporal graphs G Aydemir, AK Akan, F Güney arXiv preprint arXiv:2207.00255 , 2022 2022 Citations: 8
Stochastic video prediction with structure and motion AK Akan, S Safadoust, F Güney arXiv preprint arXiv:2203.10528 , 2022 2022 Citations: 15
Slamp: Stochastic latent appearance and motion prediction AK Akan, E Erdem, A Erdem, F Güney Proceedings of the IEEE/CVF international conference on computer vision … , 2021 2021 Citations: 62
Just noticeable difference for machines to generate adversarial images AK Akan, MA Genc, FTY Vural 2020 IEEE International Conference on Image Processing (ICIP), 1901-1905 , 2020 2020 Citations: 9
Modeling and Decoding Complex Problem Solving Process by Artificial Neural Networks AK Akan, BB Kivilcim, E Akbas, SD Newman, FTY Vural 2019 27th Signal Processing and Communications Applications Conference (SIU … , 2019 2019
MOST CITED SCHOLAR PUBLICATIONS
Adapt: Efficient multi-agent trajectory prediction with adaptation G Aydemir, AK Akan, F Güney Proceedings of the IEEE/CVF International Conference on Computer Vision … , 2023 2023 Citations: 135
Stretchbev: Stretching future instance prediction spatially and temporally AK Akan, F Güney European Conference on Computer Vision, 444-460 , 2022 2022 Citations: 72
Slamp: Stochastic latent appearance and motion prediction AK Akan, E Erdem, A Erdem, F Güney Proceedings of the IEEE/CVF international conference on computer vision … , 2021 2021 Citations: 62
Infinity-rope: Action-controllable infinite video generation emerges from autoregressive self-rollout H Yesiltepe, THS Meral, AK Akan, K Oktay, P Yanardag arXiv preprint arXiv:2511.20649 , 2025 2025 Citations: 24
Stochastic video prediction with structure and motion AK Akan, S Safadoust, F Güney arXiv preprint arXiv:2203.10528 , 2022 2022 Citations: 15
Slot-guided adaptation of pre-trained diffusion models for object-centric learning and compositional generation Y Yemez International Conference on Learning Representations 2025, 93113-93126 , 2025 2025 Citations: 9
Just noticeable difference for machines to generate adversarial images AK Akan, MA Genc, FTY Vural 2020 IEEE International Conference on Image Processing (ICIP), 1901-1905 , 2020 2020 Citations: 9
Trajectory forecasting on temporal graphs G Aydemir, AK Akan, F Güney arXiv preprint arXiv:2207.00255 , 2022 2022 Citations: 8
Just noticeable difference for machine perception and generation of regularized adversarial images with minimal perturbation AK Akan, E Akbas, FTY Vural Signal, Image and Video Processing 16 (6), 1595-1606 , 2022 2022 Citations: 6
Compositional video synthesis by temporal object-centric learning AK Akan, Y Yemez arXiv preprint arXiv:2507.20855 , 2025 2025 Citations: 2
Stochastic future prediction in real world driving scenarios AK Akan arXiv preprint arXiv:2209.10693 , 2022 2022 Citations: 1
Aligning Latent Geometry for Spherical Flow Matching in Image Generation THS Meral, K Oktay, H Yesiltepe, AK Akan, P Yanardag arXiv preprint arXiv:2605.15193 , 2026 2026
Learning Object-Centric Representations Based on Slots in Real World Scenarios AK Akan arXiv preprint arXiv:2509.24652 , 2025 2025
Modeling and Decoding Complex Problem Solving Process by Artificial Neural Networks AK Akan, BB Kivilcim, E Akbas, SD Newman, FTY Vural 2019 27th Signal Processing and Communications Applications Conference (SIU … , 2019 2019
