Skip to content

Author

Ilkay Ulusoy

2 papers indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Conference Jul 2026

Evaluating the Impact of CSI Preprocessing on WiFi-Based Human Activity Recognition

Wi-Fi Channel State Information (CSI) is a robust, privacy-preserving modality for Human Activity Recognition (HAR). Since raw CSI suffers from hardware desynchronizations and noise, preprocessing is vital. This study conducts an empirical ablation of CSI preprocessing using a fixed Two-Stream 2D CNN(Convolutional Neural Network) to quantify its impact on classification accuracy and latency. Results reveal that preprocessing, apart from architectural complexity, is the primary driver of the accuracy-latency trade-off. Computationally heavy methods like Hampel filtering introduce massive latency (>162 ms) without accuracy gains. In contrast, lightweight frequency-domain filtering consistently yields superior results. Specifically, dual-stream Butterworth bandpass filtering achieves 96.43% accuracy with only 49.08 ms latency. These findings demonstrate that isolating motion-relevant frequencies enables efficient, high-performance HAR suitable for real-time edge deployment.

Yiğit Can Çelik, Erdem Bera Emiroğlu, Sajjad Baghaee et al. · 0 citations
Conference Jul 2026

Analysis of Meta-Reinforcement Learning on Transfer Learning for HVAC Control

Deploying reinforcement learning (RL) agents for heating, ventilation, and air-conditioning (HVAC) control across climatically diverse buildings is challenging, as policies trained in one climate often fail to generalize to others. We investigate transfer and meta-reinforcement learning for building climate control using Sinergym with a custom EnergyPlus model of a real room, comparing Double Deep Q-Network (DDQN) and Model-Agnostic Meta Learning DDQN (MAML-DDQN) across three transfer scenarios on a target very hot climate. Fine-tuned DDQN reduces temperature violations by a factor of 35 over the from-scratch baseline and improves reward by 86% over zero-shot transfer, while MAML-DDQN converges in less than half the fine-tuning episodes and exhibits substantially lower variance across seeds. Results highlight a key trade-off: MAML-DDQN is preferable at adaptation-focused deployments, while fine-tuned DDQN is superior when maximum performance is the objective.

Ulaş Filiz, Sajjad Baghaee, Mubeen Ahmed Fayyaz et al. · 0 citations