Publications by authors named "Sumito Tsunegi"
Article Synopsis
- The paper explores how nanomagnet magnetization dynamics, influenced by random magnetic fields, can be related to brain-inspired computing, specifically in spintronics.
- It utilizes numerical simulations of the Thiele equation to reveal that input-driven synchronization happens despite weak perturbations and identifies new phenomena of chaotic behavior within the vortex core dynamics.
- The research also evaluates the connection between ordered and chaotic phases through the Lyapunov exponent, while further investigating how these dynamics impact the computational capabilities of physical reservoir computing.
Recent studies have shown that nonlinear magnetization dynamics excited in nanostructured ferromagnets are applicable to brain-inspired computing such as physical reservoir computing. The previous works have utilized the magnetization dynamics driven by electric current and/or magnetic field. This work proposes a method to apply the magnetization dynamics driven by voltage control of magnetic anisotropy to physical reservoir computing, which will be preferable from the viewpoint of low-power consumption.
View Article and Find Full Text PDFThe brain naturally binds events from different sources in unique concepts. It is hypothesized that this process occurs through the transient mutual synchronization of neurons located in different regions of the brain when the stimulus is presented. This mechanism of 'binding through synchronization' can be directly implemented in neural networks composed of coupled oscillators.
View Article and Find Full Text PDFArticle Synopsis
- Neuromorphic computing with spintronic devices like spin-torque oscillators (STOs) aims for energy-efficient data processing but faces challenges with stochasticity in magnetization dynamics that affect accuracy.
- *The random nature of this behavior can be beneficial for stochastic computing and machine learning, making it essential to find ways to both suppress and enhance stochastic responses in these devices.
- *Recent experiments showed that using spin-transfer effects can reduce stochasticity when transitioning between states, while applying a radio-frequency magnetic field can help minimize randomness during the oscillating state.
Physical reservoir computing is a type of recurrent neural network that applies the dynamical response from physical systems to information processing. However, the relation between computation performance and physical parameters/phenomena still remains unclear. This study reports our progress regarding the role of current-dependent magnetic damping in the computational performance of reservoir computing.
View Article and Find Full Text PDFArticle Synopsis
- The study focuses on the frequency spectrum of spin torque nano-oscillators, emphasizing the complexity introduced by thermal and flicker noise, which affects long-term stability.
- Experimental measurements were conducted on spin torque vortex oscillators, analyzing phase noise and spectrum while varying measurement time duration.
- The research combines theoretical modeling with experimental data to enhance understanding of noise impacts on oscillator stability and provides insights for improving performance in practical applications.
Article Synopsis
- The text discusses the use of reservoir computing neural networks to test neuromorphic computing hardware, specifically for tasks like automatic speech recognition.
- It highlights the significance of acoustic transformations that convert sound waves into frequency maps, which can affect how well the neuromorphic hardware performs in recognizing speech.
- The research quantifies the contributions of both the acoustic transformations and the hardware, demonstrating that the non-linearity in these transformations is crucial for effective feature extraction and that they serve as a benchmark for evaluating different neuromorphic devices.
In recent years, artificial neural networks have become the flagship algorithm of artificial intelligence. In these systems, neuron activation functions are static, and computing is achieved through standard arithmetic operations. By contrast, a prominent branch of neuroinspired computing embraces the dynamical nature of the brain and proposes to endow each component of a neural network with dynamical functionality, such as oscillations, and to rely on emergent physical phenomena, such as synchronization, for solving complex problems with small networks.
View Article and Find Full Text PDFArticle Synopsis
- Synchronized nonlinear oscillator networks are essential for applications like phased array wave generators and neuromorphic systems, with stable synchronization being a major challenge.
- This study experimentally shows that it is possible to scale synchronized spin-torque oscillator networks up to eight oscillators, resulting in increased emitted power and quality factor.
- The synchronization stability achieved lasts over 1.6 milliseconds, equivalent to 10 oscillation periods, indicating that spin-torque oscillators are promising for future applications in synchronized networks.
Article Synopsis
- Neurons in the brain function as nonlinear oscillators for processing information, inspiring advances in neuromorphic computing, which aims to replicate this behavior using nanoscale oscillators.
- Creating a chip with 10 oscillators in a small area requires each oscillator to be under one micrometer, but nanoscale devices face challenges like noise and instability that hinder reliable data processing.
- A successful experiment demonstrated that a nanoscale spintronic oscillator can recognize spoken digits with accuracy comparable to advanced neural networks, suggesting a new path for efficient, low-power on-chip computations.
The self-synchronization of spin torque oscillators is investigated experimentally by re-injecting its radiofrequency (rf) current after a certain delay time. We demonstrate that the integrated power and spectral linewidth are improved for optimal delays. Moreover by varying the phase difference between the emitted power and the re-injected one, we find a clear oscillatory dependence on the phase difference with a 2π periodicity of the frequency of the oscillator as well as its power and linewidth.
View Article and Find Full Text PDF