A5078918094- Advanced Memory and Neural Computing
- Neural Networks and Reservoir Computing
- Physics of Superconductivity and Magnetism
- Ferroelectric and Negative Capacitance Devices
- Neural dynamics and brain function
- Advanced Data Storage Technologies
- Advancements in Semiconductor Devices and Circuit Design
- Magnetic properties of thin films
- Neural Networks and Applications
- Quantum and electron transport phenomena
- Quantum Computing Algorithms and Architecture
- Semiconductor materials and devices
- VLSI and FPGA Design Techniques
- Parallel Computing and Optimization Techniques
- Low-power high-performance VLSI design
- Neuroscience and Neural Engineering
- Microgrid Control and Optimization
- Receptor Mechanisms and Signaling
- Magneto-Optical Properties and Applications
- Photovoltaic System Optimization Techniques
- Meteorological Phenomena and Simulations
University of Southern California
2022-2025
TOBB University of Economics and Technology
2016-2023
Abstract A novel high-fan-in differential superconductor neuron structure designed for ultra-high-performance Spiking Neural Network (SNN) accelerators is presented. Utilizing a allows us to design SNN with more synaptic connections, enhancing the overall network capabilities. The proposed based on electronics fabric, incorporating multiple superconducting loops, each two Josephson Junctions. This arrangement enables input data branch have positive and negative inductive coupling, supporting...
Rapid Single Flux Quantum (RSFQ) circuits are the most evolved superconductor logic family. However, need to clock each cell and deep pipeline causes a complex network with large skew. This results in lower throughput high latency RSFQ. work introduces an asynchronous RSFQ library that incorporates {\alpha}-cell, enabling bidirectional signal paths circuits. The {\alpha}-cell mitigates for by allowing reverse flow, minimizing routing, compact circuit designs. We demonstrate library's...
In five and a half years, the ColdFlux project under IARPA SuperTools program pushed boundaries of digital analog superconductor electronic design automation (S-EDA) tools. The demanded significant beyond-state-of-the-art deliverables in four main areas: RTL synthesis, architectures, verification; layout synthesis; physical test; technology CAD cell library design. Through work academic groups scattered over continents, effort forged into powerful set open-source commercial S-EDA tools...
Single Flux Quantum (SFQ) technology represents a groundbreaking advancement in computational efficiency and ultra-high-speed neuromorphic processing. The key features of SFQ technology, particularly data representation, transmission, processing through pulses, closely mirror fundamental aspects biological neural structures. Consequently, SFQ-based circuits emerge as an ideal candidate for realizing Spiking Neural Networks (SNNs). This study presents proof-of-concept demonstration SNN...
Neuromorphic computing and artificial neurons have been shown to improve the solution for some of complex problems conventional computers. We present a spiking soma Josephson junction-based (JJ-Soma) circuit that consists double-junction SQUID interfered with resistor (threshold loop), decaying superconductor loop cut by resistor, which is coupled SQUIDlike structure. The proposed has three main properties: 1) ultrahigh-speed operation minimal power consumption; 2) compatibility standard...
Due to low power consumption and high-speed performance, superconductor circuit technology has emerged as an attractive compelling post-CMOS candidate. However, the design of dense memory circuits presents a significant challenge, especially for tasks that demand substantial resources. While cells offer impressive speed, their limited density is primary yet-to-be-solved challenge. This study tackles this challenge head-on by introducing novel Non-Destructive Readout (NDRO) unit with single...
We present an on-chip trainable neuron circuit. Our proposed circuit aims at bio-inspired spike-based time-dependent data computation for training spiking neural networks (SNN). The thresholds of neurons can be increased or decreased depending on the desired application-specific spike generation rate. This mechanism is scalable and provides us with a flexible structure design. simulated under different operating scenarios thermal noise included. circuits are designed optimized MIT LL SFQ5ee...
A novel high-fan-in differential superconductor neuron structure designed for ultra-high-performance Spiking Neural Network (SNN) accelerators is presented. Utilizing a allows us to design SNN with more synaptic connections, enhancing the overall network capabilities. The proposed based on electronics fabric, incorporating multiple superconducting loops, each two Josephson Junctions. This arrangement enables input data branch have positive and negative inductive coupling, supporting...
Abstract Neural networks and neuromorphic computing represent fundamental paradigms as alternative approaches to Von-Neumann-based implementations, advancing in the applications of deep learning machine vision. Nonetheless, conventional semiconductor circuits encounter challenges achieving ultra-fast processing speed low power consumption due their dissipative properties. Conversely, single flux quantum exhibit inherent spiking behavior, showcasing characteristics a promising candidate for...
Artificial neural networks inspired by brain operations can improve the possibilities of solving complex problems more efficiently. Today's computing hardware, on other hand, is mainly based von Neumann architecture and CMOS technology, which inefficient at implementing networks. For first time, we propose an ultrahigh speed, spiking neuromorphic processor built upon single flux quantum (SFQ) artificial neurons (JJ-Neuron). Proposed has potential to provide higher performance power...
Single flux quantum (SFQ) logic family is an attractive alternative to CMOS technology with the promise of more than two–three orders magnitude improvement in energy-delay product. However, component-level parameter variations that arise during fabrication process SFQ circuits tend be very high, which means fabricated may run into major functional and/or performance issues. Therefore, optimizing cells maximize their operating margins under different variability sources required. In this...
High-performance computing that involves superconducting digital circuits is one of the promising technologies. A number groups have already demonstrated working prototypes CPUs or ALUs. However, bottlenecks these it very difficult to large memories with high speed and low-power consumption. One potential candidates compatible available foundries might enable on-chip memory vortex transitional (VTM) cell. VTM operation mainly based on dc I/O rather than single-flux quantum I/O. cell four...
The main challenge for the hardware implementation of spiking neural networks is design a reliable neuron. Soma, which nucleus neuron, key part such design. More precisely, soma must accurately capture excitatory/inhibitory interactions, intrinsic charge dynamics, refractory period, spike encoding, neuronal action potential, and output firing processes in order to mimic corresponding biological processes. This work presents an artificial cell with excitatory inhibitory inputs, called...
Superconductor electronics (SCE) is a promising complementary and beyond CMOS technology. However, despite its practical benefits, the realization of SCE logic faces significant challenge due to absence dense scalable nonvolatile memory designs. While various technologies, including Non-destructive readout, vortex transitional (VTM), magnetic memory, have been explored, achieving superconductor random-access (RAM) crossbar array remains challenging. This paper introduces novel, nonvolatile,...
Abstract Despite superconductor electronics (SCE) advantages, the realization of SCE logic faces a significant challenge due to absence dense and scalable nonvolatile memory. While various memory technologies, including Non-destructive readout (NDRO), vortex transitional (VTM), magnetic memory, have been explored, designing crossbar array achieving random-access (RAM) remains challenging. This work introduces novel, nonvolatile, high-density, vortex-based design for logic, called bistable...
Artificial neurons provide a new way of computation for neuro-inspired algorithms, and the abilities may efficiently solve challenges. We propose implementations logic gates ( <sc xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">and, or, xor</small> , Majority), full adder, subtractor, even parity generator, 2-bit multiplier circuit formed by Josephson junction-based soma (JJ-Soma) standard Rapid Single Flux Quantum (SFQ) digital library cells. The...
Single flux quantum (SFQ) technology has garnered significant attention due to its low switching power and high operational speed. Researchers have been actively pursuing more advanced devices technologies further reduce the reliance on inductors, bias, dynamic power. Recently, innovative magnetic Josephson junction emerged, enhancing field of superconductor electronics (SCE) logic. This paper introduces a novel cell library design that relies entirely junctions (JJs), showing promising...
We present an on-chip trainable neuron circuit. Our proposed circuit suits bio-inspired spike-based time-dependent data computation for training spiking neural networks (SNN). The thresholds of neurons can be increased or decreased depending on the desired application-specific spike generation rate. This mechanism provides us with a flexible design and scalable structure. demonstrate structure under different operating scenarios. circuits are designed optimized MIT LL SFQ5ee fabrication...
Neural networks and neuromorphic computing play pivotal roles in deep learning machine vision. Due to their dissipative nature inherent limitations, traditional semiconductor-based circuits face challenges realizing ultra-fast low-power neural networks. However, the spiking behavior characteristic of single flux quantum (SFQ) positions them as promising candidates for (SNNs). Our previous work showcased a JJ-Soma design capable operating at tens gigahertz while consuming only fraction power...