Reservoir computing is a computational framework suited for temporal/sequential data processing. It is derived from several recurrent neural network models, including echo state networks and liquid state machines. A reservoir computing system consists of a reservoir for mapping inputs into a high-dimensional space and a readout for pattern analysis from the high-dimensional states in the reservoir. The reservoir is fixed and only the readout is trained with a simple method such as linear regression and classification. Thus, the major advantage of reservoir computing compared to other recurrent neural networks is fast learning, resulting in low training cost. Another advantage is that the reservoir without adaptive updating is amenable to hardware implementation using a variety of physical systems, substrates, and devices. In fact, such physical reservoir computing has attracted increasing attention in diverse fields of research. The purpose of this review is to provide an overview of recent advances in physical reservoir computing by classifying them according to the type of the reservoir. We discuss the current issues and perspectives related to physical reservoir computing, in order to further expand its practical applications and develop next-generation machine learning systems.

The nonlinear dynamics of rotating low m (poloidal mode number) tearing modes in a tokamak with external resonant magnetic perturbations is examined. Nonlinear evolution equations for the island width and the toroidal rotation frequency are derived within the two-fluid magnetohydrodynamic model, taking into account the plasma rotation and neoclassical parallel viscosity. The nonlinear stability of magnetic islands interacting with a static external magnetic perturbation is considered, and the critical magnetic field for the appearance of a locked mode is determined. It is shown that the coupling of the perpendicular and longitudinal plasma flow due to the neoclassical plasma viscosity enhances the amplitude of the critical magnetic field compared to the value obtained in a slab approximation. The perpendicular plasma viscosity causes a finite phase shift between the applied external field and the magnetic island, and further increases the value of the critical magnetic field required to induce a magnetic island. 0 1995 American Institute of Ph.ysics.

No abstract

No abstract

Transgenic mice with vascular endothelial growth factor (VEGF) driven by the rhodopsin promoter (rho/VEGF mice) develop neovascularization that originates from the deep capillary bed of the retina and grows into the subretinal space. In rho/VEGF mice, VEGF expression in photoreceptors begins between postnatal days 5 and 7, the period when the deep capillary bed is developing. An important question is whether or not the developmental stage of the deep capillary bed is critical for occurrence of neovascularization. Also, although rho/VEGF mice are extremely useful for the study of ocular neovascularization, there are some applications for which the early onset of VEGF expression is a disadvantage. In this study, we used the reverse tetracycline transactivator (rtTA) inducible promoter system coupled to either the rhodopsin or interphotoreceptor retinoidbinding protein ( Damage to retinal capillaries resulting in retinal ischemia is a constant feature in diseases complicated by retinal neovascularization. Therefore, the demonstration that vascular endothelial cell growth factor (VEGF) is up-regulated by hypoxia 1,2 focused attention on VEGF as a candidate mediator of retinal neovascularization. Circumstantial evidence suggesting that VEGF plays a role was provided by temporal and spatial correlation of VEGF expression with retinal neovascularization in patients and animal models. 3-8 Inhibition of retinal neovascularization by several different types of VEGF antagonists has demonstrated that VEGF is a necessary stimulatory factor. 9 -13