Browsing by Author "Elahi, Sepehr"

Now showing 1 - 7 of 7

Restricted
Bardezag orphanage
(Bilkent University, 2018) Khan, Muhammad Arham; Khalid, Muhammad Bilal Bin; Fakhouri, Mohamad; Aftab, Abdül Moiz; Elahi, Sepehr
Bu makalede, Bardezag Yetimhanesi’nin İzmit bölgesinde açılmasına neden olan Ermeni ve Osmanlı Devleti arasında gerçekleşen çatışmalar ve Ermeni ailelerin çocuklarını yetim ve güçsüz bırakan olaylar incelenmiştir. Bu olayların başta gelenlerinden bazıları Ermenilerin toplu olarak katledilmesi, bölgede yaşayan Ermenilerin ödediği vergilerin ikiye çıkarılması ve Türklerin Ermenilere herhangi bir yardım ve destekte bulunmaması olmuştur. Yetimhanenin kurulmasını sağlayan nedenler tartışıldıktan sonra yetimhaneye yabancı ülkeler tarafından yapılan bağışların hızla artan yetim sayısına yetmemesinin ardından giderlerini hangi yollarla karşıladığı incelenmiş ve çocuklar arasında iş bölümü yaparak giderlerini en aza indirdiği görülmüştür.
Open Access
Feedback adaptive learning for medical and educational application recommendation
(IEEE, 2020) Tekin, Cem; Elahi, Sepehr; Van Der Schaar, M.
Recommending applications (apps) to improve health or educational outcomes requires long-term planning and adaptation based on the user feedback, as it is imperative to recommend the right app at the right time to improve engagement and benefit. We model the challenging task of app recommendation for these specific categories of apps-or alike-using a new reinforcement learning method referred to as episodic multi-armed bandit (eMAB). In eMAB, the learner recommends apps to individual users and observes their interactions with the recommendations on a weekly basis. It then uses this data to maximize the total payoff of all users by learning to recommend specific apps. Since computing the optimal recommendation sequence is intractable, as a benchmark, we define an oracle that sequentially recommends apps to maximize the expected immediate gain. Then, we propose our online learning algorithm, named FeedBack Adaptive Learning (FeedBAL), and prove that its regret with respect to the benchmark increases logarithmically in expectation. We demonstrate the effectiveness of FeedBAL on recommending mental health apps based on data from an app suite and show that it results in a substantial increase in the number of app sessions compared with episodic versions of ϵn -greedy, Thompson sampling, and collaborative filtering methods.
Embargo
High-precision laser focus positioning of rough surfaces by deep learning
(Elsevier Ltd, 2023-05-18) Polat, Can; Yapici, Gizem Nuran; Elahi, Sepehr; Elahi, Parviz
This work presents a precise positioning detection based on a convolutional neural network (CNN) to control the laser focus in laser material processing systems. The images of the diffraction patterns measured at different positions of the laser focus concerning the workpiece are classified in the range of the Rayleigh length of the focusing lens with an increment of about 7% of it. The experiment was carried out on different materials with different levels of surface roughness, such as copper, silicon, and steel, and over 99% accuracy in the positioning detection was achieved. Considering surface roughness and camera noise, a theoretical model is established, and the effects of these parameters on the accuracy of focus detection are also presented. The proposed method exhibits a noise-robust focus detection system and the potential for many precise positioning detection systems in industry and biology. © 2023 Elsevier Ltd.
Open Access
Machine learning-based high-precision and real-time focus detection for laser material processing systems
(S P I E - International Society for Optical Engineering, 2022-05-17) Polat, Can; Yapıcı, Gizem Nuran; Elahi, Sepehr; Elahi, Parviz
This work explores a real-time and high precision focus finding for the ultrafast laser material processing for a different types of materials. Focus detection is essential for laser machining because an unfocused beam cannot affect the material and, at worst, a destructive effect. Here, we compare CNN and non-CNN-based approaches to focus detection, ultimately proposing a robust CNN model that can achieve high performance when only trained on a portion of the dataset. We use an ordinary lens (11 mm focal length, 0.25 NA) and a CMOS camera. Our robust CNN model achieved a focus prediction accuracy of 95% when identifying focus distances in -150, -140,...,0,...,150 µm, each step is about 7% of the Rayleigh length, and a high processing speed of 1000+ Hz on a CPU.
Open Access
Noise robust focal distance detection in laser material processing using CNNs and Gaussian processes
(S P I E - International Society for Optical Engineering, 2022-05-17) Elahi, Sepehr; Polat, Can; Safarzadeh, Omid; Elahi, Parviz
In this work, we investigate the effects of noise on real-time focal distance control for laser material processing by generating the images of a sample at different focal lengths using Fourier optics and then designing, training, and testing a deep learning model in order to detect the focal distances from the simulated images with varying standard deviations of added noise. We simulate both input noise, such as noise due to surface roughness, and output noise, such as detection camera noise, by adding zero-mean Gaussian noise to the source wave and the simulated image, respectively, for different focal distances. We then train a convolutional neural network combined with a Gaussian process classifier to predict focus distances of noisy images together with confidence ratings for the predictions.
Open Access
Online context-aware task assignment in mobile crowdsourcing via adaptive discretization
(IEEE, 2022-09-22) Elahi, Sepehr; Nika, Andi; Tekin, Cem
Mobile crowdsourcing is rapidly boosting the Internet of Things revolution. Its natural development leads to an adaptation to various real-world scenarios, thus imposing a need for wide generality on data-processing and task-assigning methods. We consider the task assignment problem in mobile crowdsourcing while taking into consideration the following: (i) we assume that additional information is available for both tasks and workers, such as location, device parameters, or task parameters, and make use of such information; (ii) as an important consequence of the worker-location factor, we assume that some workers may not be available for selection at given times; (iii) the workers' characteristics may change over time. To solve the task assignment problem in this setting, we propose Adaptive Optimistic Matching for Mobile Crowdsourcing (AOM-MC), an online learning algorithm that incurs O~(T(D¯+1)/(D¯+2)+ϵ) regret in T rounds, for any ϵ>0 , under mild continuity assumptions. Here, D¯ is a notion of dimensionality which captures the structure of the problem. We also present extensive simulations that illustrate the advantage of adaptive discretization when compared with uniform discretization, and a time- and location-dependent crowdsourcing simulation using a real-world dataset, clearly demonstrating our algorithm's superiority to the current state-of-the-art and baseline algorithms.
Open Access
Real-time image-based droplet measurement
(Chemical and Biological Microsystems Society, 2020) Elahi, Sepehr; Kalantarifard, Ali; Kalantarifard, Fatemeh; Elbüken, Çağlar
The ability to measure physical properties of droplets in real-time is required to design precise operations on droplet-based systems. In this study, we implemented a real-time droplet tracker that tracks the positions of droplets and measures droplet generation frequency as well as droplets' physical properties, such as size, size distribution, shape, velocity, circularity. Furthermore, using the droplet length, we use curve fitting to determine the dispersed phase viscosity. Our droplet tracker is implemented in Python, using the OpenCV library and can be run on a routine PC.