A5015017700- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Robotic Path Planning Algorithms
- Advanced Neural Network Applications
- Parallel Computing and Optimization Techniques
- Robotics and Sensor-Based Localization
- UAV Applications and Optimization
- Experimental Learning in Engineering
- Quantum Computing Algorithms and Architecture
- Cloud Computing and Resource Management
- Modular Robots and Swarm Intelligence
- Augmented Reality Applications
- Teaching and Learning Programming
- Power Line Communications and Noise
- Advanced MEMS and NEMS Technologies
- Physical Activity and Health
- Domain Adaptation and Few-Shot Learning
- Air Traffic Management and Optimization
- Sensor Technology and Measurement Systems
- Cloud Computing and Remote Desktop Technologies
- Gender and Technology in Education
- Multimodal Machine Learning Applications
- Advanced Aircraft Design and Technologies
- Mechatronics Education and Applications
- Electromagnetic Compatibility and Noise Suppression
University of California, Berkeley
2019-2024
Marmara University
2020-2021
Berkeley College
2021
The University of Texas at Austin
2017-2019
Mi̇lli̇ Eği̇ti̇m Bakanliği
2017
Sakarya University
2012-2017
DNN accelerators are often developed and evaluated in isolation without considering the cross-stack, system-level effects real-world environments. This makes it difficult to appreciate impact of Systemon-Chip (SoC) resource contention, OS overheads, programming-stack inefficiencies on overall performance/energy-efficiency. To address this challenge, we present Gemmini, an open-source, full-stack accelerator generator. Gemmini generates a wide design-space efficient ASIC from flexible...
Unmanned Aerial Vehicles (UAVs) are getting closer to becoming ubiquitous in everyday life. Among them, Micro (MAVs) have seen an outburst of attention recently, specifically the area with a demand for autonomy. A key challenge standing way making MAVs autonomous is that researchers lack comprehensive understanding how performance, power, and computational bottlenecks affect MAV applications. must operate under stringent power budget, which severely limits their flight endurance time. As...
High-performance kernel libraries are critical to exploiting accelerators and specialized instructions in many applications. Because compilers difficult extend support diverse rapidly-evolving hardware targets, automatic optimization is often insufficient guarantee state-of-the-art performance, these commonly still coded optimized by hand, at great expense, low-level C assembly. To better development of high-performance for hardware, we propose a new programming language, Exo, based on the...
The Internet of Things (IoT) is rapidly enabling applications in many different fields by embedding itself into the physical world. Many potential IoT devices require some level machine learning or cognitive capability to be truly effective, but high computational complexity algorithms makes them unsuitable for low-power processors. To understand design challenges devices, authors study a drone application space called Flying IoT. They find that computing provided typical processors falls...
Driven by the wide adoption of deep neural networks (DNNs) across different application domains, multi-tenancy execution, where multiple DNNs are deployed simultaneously on same hardware, has been proposed to satisfy latency requirements applications while improving overall system utilization. However, execution could lead undesired system-level resource contention, causing quality-of-service (QoS) degradation for latency-critical applications. To address this challenge, we propose MoCA, an...
This work presents a 16mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> heterogeneous RISC-V system-on-a-chip (SoC) composed of high-performance out-of-order core, energy-efficient in-order data-parallel vector accelerator, and systolic array deep neural network (DNN) accelerator in low-power Intel 22FFL for general-purpose compute, DNN, workloads. The SoC is fully open-source components, including second-generation Berkeley...
Abstract Determining input impedance, assigning reflection coefficient, and matching load impedance are important transmission line problems those involve complex calculation steps because of connection architectures in lines. Efficient understanding such undergraduate graduate courses should be supported by rich visualization tools. The Smith chart is a convenient tool for calculations especially wave propagation maximum power theory that discusses problems. In this context, we can say Java...
There is growing interest in object detection advanced driver assistance systems and autonomous robots vehicles. To enable such innovative systems, we need faster detection. In this work, investigate the trade-off between accuracy speed with domain-specific approximations, i.e. category-aware image size scaling proposals scaling, for two state-of-the-art deep learning-based meta-architectures. We study effectiveness of applying approximation both statically dynamically to understand...
Abstract Quantum computation based on quantum physics is spreading rapidly with many interdisciplinary studies from machine learning to genetics. After the unveiling of real computers, this process has been rising almost exponentially. We are about experience an effect computers similar created by first electromechanical very large‐scale integrated circuit computers. In programming we in a place classical at early stage. There various models computing, and most widely used model gates....
Autonomous and mobile cyber-physical machines are becoming an inevitable part of our future. In particular, Micro Aerial Vehicles (MAVs) have seen a resurgence in activity. With multiple use cases, such as surveillance, search rescue, package delivery, more, these unmanned aerial systems on the cusp demonstrating their full potential. Despite promises, face many challenges, one most prominent which is low endurance caused by limited onboard energy. Since success mission depends whether drone...
The realization of effective physics applications is thought as a privileged phenomenon due to the lack transferring basic laws produce products. In addition, development some depending on theory causes discussion and questioning scholars competences which must be suitable for life. Assessment made with undergraduates shows that difficulties are observed embodiment problems in engineering technical sciences. There lot licensed softwares analyzing complex physical fields. But these programs...
Self-attention matrices of Transformers are often highly sparse because the relevant context each token is typically limited to just a few other tokens in sequence. To reduce computational burden self-attention on Transformer inference, we propose static, structured, attention masks that split into dense regions, skipping computations outside these regions while reducing inside regions. support proposed mask structure, design an entropy-aware finetuning algorithm naturally encourage sparsity...
Quantum computers are expected to offer an effect similar the influence of early integrated circuit computers. According current systems, it is predicted that they will play effective role in emergence a stronger technology with increasing speed. Some high-tech companies have quantum computer designs actively put into use from research and development phase problem-solving these different architectures. Unlike others, IBM launched cloud-based software infrastructure 2016 first introduced its...
As educators are increasingly using and developing courseware to use with their students, it gets more important understand how high quality should be designed, developed used. In this study, based on idea, a web instructional design for Electronics course was the vocational school Department grade 2. purpose is present development evaluation process of support creation designs. The study survey model. A scale called "Technical Pedagogical Bases Web Based Instruction" by researchers assess...
Quantum computers are becoming more and powerful practical each day with the perfection of coherence time superposition states in terms fidelity, error correction, power dissipation, noise. have been used successfully solution non-deterministic polynomial-time problems for a while. Using quantum algorithms to find ways speed up computing some hard reveals concept called supremacy. With improvement decrease rates, companies that develop offered cloud-based platforms make their publicly...
Physical phenomena and the nature of materials require both qualitative quantitative research to be understood thoroughly.This process often necessitates use complicated expensive equipments.Also in electrical-electronics discipline, especially semiconductor behavior them have a special place.In engineering education, it is more important for undergraduates understand material results that at basic level.For this purpose sufficient analyze with interactive software unlike physics scholars...
Autonomous-mobile cyber-physical machines are part of our future. Specifically, unmanned-aerial-vehicles have seen a resurgence in activity with use-cases such as package delivery. These systems face many challenges their low-endurance caused by limited onboard-energy, hence, improving the mission-time and energy importance. Such improvements traditionally delivered through better algorithms. But premise is that more powerful efficient onboard-compute should also address problem. This paper...
Elektrik-Elektronik muhendisliginde, karmasik devre problemleri lisans ogrencilerinin algilamada gucluk cektikleri bir konudur. Ogretim teknolojileri kullanilarak ve soyut kavramlar benzetim canlandirmalarla somutlastirilarak ogrencilerin derslerdeki basarisi arttirilabilir. Ilgili disiplinle iliskili problemlerin ogrenciler icin somutlastirilmasi, konulari kalici sekilde ogrenmesine yardimci olacaktir. Karmasik formullerin duragan ezberletilmesi yerine, bu yazilim uygulamalari yaptirarak...
Elektrik-Elektronik mühendisliğinin iki alt teorisinden biri olan alan teorisi lisans öğrencilerinin algılamada güçlük çektikleri bir konudur. Lisans döneminde elektromanyetik alanlar kapsamında teorisi, parçalı diferansiyel denklemler ve integral yöntemlerinin kullanıldığı analitik yöntemlerle anlatılmaktadır. Alan problemlerinin matematiksel modelinin çıkartılması esasına dayalı çözüm yöntemleri matematik fizik altyapılarına bağlı olarak algılama güçlüklerinin yaşanmasına neden...