Researcher:
Yılmaz, Mustafa Akın

Profile Picture
ORCID

Job Title

PhD Student

First Name

Mustafa Akın

Last Name

Yılmaz

Name

Name Variants

Yılmaz, Mustafa Akın

Email Address

Birth Date

Filters

Advanced Search

Filter by

Settings

Researcher Of Publications: search.filters.isAuthorOfPublication.69402f65-99ad-4471-b8e5-c155b61e9ec6

Search Results

Now showing 1 - 10 of 15
  • Placeholder
    PublicationMetadata only
    Flexible-rate learned hierarchical bi-directional video compression with motion refinement and frame-level bit allocation
    (IEEE Computer Society, 2022) Department of Electrical and Electronics Engineering; N/A; N/A; Tekalp, Ahmet Murat; Yılmaz, Mustafa Akın; Çetin, Eren; Faculty Member; PhD Student; Undergraduate Student; Department of Electrical and Electronics Engineering; College of Engineering; Graduate School of Sciences and Engineering; College of Engineering; 26207; N/A; N/A
    This paper presents improvements and novel additions to our recent work on end-to-end optimized hierarchical bidirectional video compression [1] to further advance the state-of-the-art in learned video compression. As an improvement, we combine motion estimation and prediction modules and compress refined residual motion vectors for improved rate-distortion performance. As novel addition, we adapted the gain unit proposed for image compression to flexible-rate video compression in two ways: first, the gain unit enables a single encoder model to operate at multiple rate-distortion operating points; second, we exploit the gain unit to control bit allocation among intra-coded vs. bi-directionally coded frames by fine tuning corresponding models for truly flexible-rate learned video coding. Experimental results demonstrate that we obtain state-of-the-art rate-distortion performance exceeding those of all prior art in learned video coding.
  • Placeholder
    PublicationMetadata only
    A new characterization approach to study the mechanical behavior of silicon nanowires
    (Springer, 2021) Esfahani, Mohammad Nasr; Taşdemir, Zuhal; Wollschlaeger, Nicole; Li, XueFei; Li, Taotao; Leblebici, Yusuf; N/A; N/A; Department of Mechanical Engineering; Zarepakzad, Sina; Yılmaz, Mustafa Akın; Alaca, Burhanettin Erdem; PhD Student; PhD Student; Faculty Member; Department of Mechanical Engineering; N/A; N/A; Koç University Surface Science and Technology Center (KUYTAM) / Koç Üniversitesi Yüzey Teknolojileri Araştırmaları Merkezi (KUYTAM); Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; N/A; 115108
    This work proposes a new approach to characterize the mechanical properties of nanowires based on a combination of nanomechanical measurements and models. Silicon nanowires with a critical dimension of 90 nm and a length of 8 mu m obtained through a monolithic process are characterized through in-situ three-point bending tests. A nonlinear nanomechanical model is developed to evaluate the mechanical behavior of nanowires. In this model, the intrinsic stress and surface parameters are examined based on Raman spectroscopy measurements and molecular dynamics simulations, respectively. This work demonstrates a new approach to measure the mechanical properties of Si nanowires by considering the surface effect and intrinsic stresses. The presented technique can be used to address the existing discrepancies between numerical estimations and experimental measurements on the modulus of elasticity of silicon nanowires.
  • Placeholder
    PublicationMetadata only
    Generating robot/agent backchannels during a storytelling experiment
    (Institute of Electrical and Electronics Engineers (IEEE), 2009) Al Moubayed, S.; Baklouti, M.; Chetouani, M.; Dutoit, T.; Mahdhaoui, A.; Martin, J. -C.; Ondas, S.; Pelachaud, C.; Urbain, J.; Department of Mechanical Engineering; Yılmaz, Mustafa Akın; Tekalp, Ahmet Murat; PhD Student; Department of Mechanical Engineering; Graduate School of Sciences and Engineering; N/A
    This work presents the development of a real-time framework for the research of Multimodal Feedback of Robots/Talking Agents in the context of Human Robot Interaction (HRI) and Human Computer Interaction (HCI). For evaluating the framework, a Multimodal corpus is built (ENTERFACE_STEAD), and a study on the important multimodal features was done for building an active Robot/Agent listener of a storytelling experience with Humans. The experiments show that even when building the same reactive behavior models for Robot and Talking Agents, the interpretation and the realization of the behavior communicated is different due to the different communicative channels Robots/Agents offer be it physical but less-human-like in Robots, and virtual but more expressive and human-like in Talking agents.
  • Placeholder
    PublicationMetadata only
    Superplastic behavior of silica nanowires obtained by direct patterning of silsesquioxane-based precursors
    (Iop Publishing Ltd, 2017) Wollschlaeger, Nicole; Oesterle, Werner; Leblebici, Yusuf; N/A; N/A; Department of Mechanical Engineering; Yılmaz, Mustafa Akın; Esfahani, Mohammad Nasr; Alaca, Burhanettin Erdem; PhD Student; PhD Student; Faculty Member; Department of Mechanical Engineering; College of Engineering / Koç University Surface Science and Technology Center (KUYTAM) / Koç Üniversitesi Yüzey Teknolojileri Araştırmaları Merkezi (KUYTAM); Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; N/A; 115108
    Silica nanowires spanning 10 mu m-deep trenches are fabricated from different types of silsesquioxane-based precursors by direct e-beam patterning on silicon followed by release through deep reactive ion etching. Nanowire aspect ratios as large as 150 are achieved with a critical dimension of about 50 nm and nearly rectangular cross-sections. In situ bending tests are carried out inside a scanning electron microscope, where the etch depth of 10 mu m provides sufficient space for deformation. Silica NWs are indeed observed to exhibit superplastic behavior without fracture with deflections reaching the full etch depth, about two orders of magnitude larger than the nanowire thickness. A large-deformation elastic bending model is utilized for predicting the deviation from the elastic behavior. The results of forty different tests indicate a critical stress level of 0.1-0.4 GPa for the onset of plasticity. The study hints at the possibility of fabricating silica nanowires in a monolithic fashion through direct e-beam patterning of silsesquioxane-based resins. The fabrication technology is compatible with semiconductor manufacturing and provides silica nanowires with a very good structural integrity.
  • Placeholder
    PublicationMetadata only
    Effect of architectures and training methods on the performance of learned video frame prediction
    (IEEE, 2019) N/A; Department of Electrical and Electronics Engineering; Yılmaz, Mustafa Akın; Tekalp, Ahmet Murat; PhD Student; Faculty Member; Department of Electrical and Electronics Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; 26207
    We analyze the performance of feedforward vs. recurrent neural network (RNN) architectures and associated training methods for learned frame prediction. To this effect, we trained a residual fully convolutional neural network (FCNN), A convolutional RNN (CRNN), and a convolutional long short-term memory (CLSTM) network for next frame prediction using the mean square loss. We performed both stateless and stateful training for recurrent networks. Experimental results show that the residual FCNN architecture performs the best in terms of peak signal to noise ratio (PSNR) at the expense of higher training and test (inference) computational complexity. the CRNN can be trained stably and very efficiently using the stateful truncated backpropagation through time procedure, and it requires an order of magnitude less inference runtime to achieve near real-time frame prediction with an acceptable performance.
  • Placeholder
    PublicationMetadata only
    Video frame prediction via deep learning
    (IEEE, 2020) N/A; Department of Electrical and Electronics Engineering; Yılmaz, Mustafa Akın; Tekalp, Ahmet Murat; PhD Student; Faculty Member; Department of Electrical and Electronics Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; 26207
    This paper provides new results over our previous work presented in ICIP 2019 on the performance of learned frame prediction architectures and associated training methods. More specifically, we show that using an end-to-end residual connection in the fully convolutional neural network (FCNN) provides improved performance. in order to provide comparative results, we trained a residual FCNN, A convolutional RNN (CRNN), and a convolutional long-short term memory (CLSTM) network for next frame prediction using the mean square loss. We performed both stateless and stateful training for recurrent networks. Experimental results show that the residual FCNN architecture performs the best in terms of peak signal to noise ratio (PSNR) at the expense of higher training and test (inference) computational complexity. the CRNN can be stably and efficiently trained using the stateful truncated backpropagation through time procedure, and requires an order of magnitude less inference runtime to achieve an acceptable performance in near real-time.
  • Placeholder
    PublicationMetadata only
    Monolithic technology for silicon nanowires in high-topography architectures
    (Elsevier, 2017) Wollschlager, Nicole; Rangelow, Ivo W.; Leblebici, Yusuf; Department of Mechanical Engineering; Esfahani, Mohammad Nasr; Yılmaz, Mustafa Akın; Alaca, Burhanettin Erdem; PhD Student; PhD Student; Faculty Member; Department of Mechanical Engineering; Koç University Surface Science and Technology Center (KUYTAM) / Koç Üniversitesi Yüzey Teknolojileri Araştırmaları Merkezi (KUYTAM); Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; N/A; 115108
    Integration of silicon nanowires (Si NWs) in three-dimensional (3D) devices including integrated circuits (ICs) and microelectromechanical systems (MEMS) leads to enhanced functionality and performance in diverse applications. The immediate challenge to the extensive use of Si NWs in modern electronic devices is their integration with the higher-order architecture. Topography-related limits of integrating Si NWs in the third dimension are addressed in this work. Utilizing a well-tuned combination of etching and protection processes, Si NWs are batch-produced in bulk Si with an extreme trench depth of 40 gm, the highest trench depth obtained in a monolithic fashion within the same Si crystal so far. The implications of the technique for the thick silicon-on-insulator (S01) technology are investigated. The process is transferred to SOI wafers yielding Si NWs with a critical dimension of 100 nm along with a trench aspect ratio of 50. Electrical measurements verify the prospect of utilizing such suspended Si NWs spanning deep trenches as versatile active components in ICs and MEMS. Introducing a new monolithic approach to obtaining Si NWs and the surrounding higher-order architecture within the same SOI wafer, this work opens up new possibilities for modem sensors and power efficient ICs. (C) 2017 Elsevier B.V. All rights reserved.
  • Placeholder
    PublicationMetadata only
    End-to-End Rate-Distortion Optimization for Bi-Directional Learned Video Compression
    (Ieee, 2020) N/A; Department of Electrical and Electronics Engineering; Yılmaz, Mustafa Akın; Tekalp, Ahmet Murat; PhD Student; Faculty Member; Department of Electrical and Electronics Engineering; Graduate School of Sciences and Engineering; College of Engineering; N/A; 26207
    Conventional video compression methods employ a linear transform and block motion model, and the steps of motion estimation, mode and quantization parameter selection, and entropy coding are optimized individually due to combinatorial nature of the end-to-end optimization problem. Learned video compression allows end-to-end rate-distortion optimized training of all nonlinear modules, quantization parameter and entropy model simultaneously. While previous work on learned video compression considered training a sequential video codec based on end-to-end optimization of cost averaged over pairs of successive frames, it is well-known in conventional video compression that hierarchical, bi-directional coding outperforms sequential compression. In this paper, we propose for the first time end-to-end optimization of a hierarchical, bi-directional motion compensated learned codec by accumulating cost function over fixed-size groups of pictures (GOP). Experimental results show that the rate-distortion performance of our proposed learned bi-directional GOP coder outperforms the state-of-the-art end-to-end optimized learned sequential compression as expected.
  • Placeholder
    PublicationMetadata only
    A numerical simulation for the stress effect in flexural micro/nano electromechanical resonators
    (American Scientific Publishers, 2015) N/A; N/A; N/A; Department of Mechanical Engineering; Yılmaz, Mustafa Akın; Esfahani, Mohammad Nasr; Biçer, Mahmut; Alaca, Burhanettin Erdem; PhD Student; PhD Student; Researcher; Faculty Member; Department of Mechanical Engineering; Graduate School of Sciences and Engineering; Graduate School of Sciences and Engineering; N/A; College of Engineering; N/A; N/A; N/A; 115108
    Resonance frequencies and quality factors of micro/nano electromechanical resonators are known to differ significantly from target values in the presence of intrinsic stresses. This stress effect is modeled for a two-port system with electrostatic actuation and capacitive read-out. A methodology is proposed to compute equivalent electrical parameters for a double-clamped beam resonator under stress. The model is verified with finite element analysis, and a number of case studies are conducted in addition. Increase in resonance frequency with increasing intrinsic tensile stress is observed under mechanical and electrical effects, while a deterioration of quality factor is evident in cases with pronounced parasitic effects. Related challenges associated with the transition to the nanoscale are computationally captured. Finally, a short formulation is provided with relevant error margins for the direct estimation of equivalent circuit parameters. The proposed approach serves as a useful tool for layout design, where all involved dimensions are considered in addition to operational variables such as bias voltage and unloaded quality factor.
  • Thumbnail Image
    PublicationOpen Access
    Self-organized variational autoencoders (self-vae) for learned image compression
    (Institute of Electrical and Electronics Engineers (IEEE), 2021) Malik, J.; Kıranyaz S.; Department of Electrical and Electronics Engineering; Tekalp, Ahmet Murat; Keleş, Onur; Yılmaz, Mustafa Akın; Güven, Hilal; Faculty Member; Department of Electrical and Electronics Engineering; College of Engineering; Graduate School of Sciences and Engineering; 26207; N/A; N/A; N/A
    In end-to-end optimized learned image compression, it is standard practice to use a convolutional variational autoencoder with generalized divisive normalization (GDN) to transform images into a latent space. Recently, Operational Neural Networks (ONNs) that learn the best non-linearity from a set of alternatives, and their “self-organized” variants, Self-ONNs, that approximate any non-linearity via Taylor series have been proposed to address the limitations of convolutional layers and a fixed nonlinear activation. In this paper, we propose to replace the convolutional and GDN layers in the variational autoencoder with self-organized operational layers, and propose a novel self-organized variational autoencoder (Self-VAE) architecture that benefits from stronger non-linearity. The experimental results demonstrate that the proposed Self-VAE yields improvements in both rate-distortion performance and perceptual image quality.