Perceptual learning, roving and the unsupervised bias

Perceptual learning improves perception through training. Perceptual learning improves with most stimulus types but fails when . certain stimulus types are mixed during training (roving). This result is surprising because classical supervised and unsupervised neural network models can cope easily with roving conditions. What makes humans so inferior compared to these models? As experimental and conceptual work has shown, human perceptual learning is neither supervised nor unsupervised but reward-based learning. Reward-based learning suffers from the so-called unsupervised bias, i.e., to prevent synaptic " drift" , the . average reward has to be exactly estimated. However, this is impossible when two or more stimulus types with different rewards are presented during training (and the reward is estimated by a running average). For this reason, we propose no learning occurs in roving conditions. However, roving hinders perceptual learning only for combinations of similar stimulus types but not for dissimilar ones. In this latter case, we propose that a critic can estimate the reward for each stimulus type separately. One implication of our analysis is that the critic cannot be located in the visual system. © 2011 Elsevier Ltd.

Design and optimization of a 250nm SOI LDMOSFET

This work is aimed at optimising the static performance of a high voltage SOI LDMOSFET. Starting with a conventional LDMOSFET, 2D and 3D numerical simulation models, able to accurately match datasheet values, have been developed. Moving from the original device, several design techniques have been investigated with the target of improving the breakdown voltage and the ON-state resistance. The considered design techniques are based on the modification of the doping profile of the drift region and the Superjunction design technique. The paper shows that a single step doping within the drift region is the best design choice for the considered device and is found to give a 24% improvement in the breakdown voltage and a 17% reduction of the ON-state resistance. © 2011 IEEE.

Low on-resistance SOI dual-trench-gate MOSFET

A low specific on-resistance (R-{{\rm on}, {\rm sp}}) integrable silicon-on-insulator (SOI) MOSFET is proposed, and its mechanism is investigated by simulation. The SOI MOSFET features double trenches and dual gates (DTDG SOI): an oxide trench in the drift region, a buried gate inset in the oxide trench, and another trench gate (TG) extended to a buried oxide layer. First, the dual gates form dual conduction channels, and the extended gate widens the vertical conduction area; both of which sharply reduce R-{{\rm on}, {\rm sp}}. Second, the oxide trench folds the drift region in the vertical direction, resulting in a reduced device pitch and R-{{\rm on}, {\rm sp}}. Third, the oxide trench causes multidirectional depletion. This not only enhances the reduced surface field effect and thus reshapes the electric field distribution but also increases the drift doping concentration, leading to a reduced R-{{\rm on}, {\rm sp}} and an improved breakdown voltage (BV). Compared with a conventional SOI lateral Double-diffused metal oxide semiconductor (LDMOS), the DTDG MOSFET increases BV from 39 to 92 V at the same cell pitch or decreases R-{{\rm on}, { \rm sp}} by 77% at the same BV by simulation. Finally, the TG extended synchronously acts as an isolation trench between the high/low-voltage regions in a high-voltage integrated circuit, saving the chip area and simplifying the isolation process. © 2006 IEEE.

Paradoxical evidence integration in rapid decision processes

Decisions about noisy stimuli require evidence integration over time. Traditionally, evidence integration and decision making are described as a one-stage process: a decision is made when evidence for the presence of a stimulus crosses a threshold. Here, we show that one-stage models cannot explain psychophysical experiments on feature fusion, where two visual stimuli are presented in rapid succession. Paradoxically, the second stimulus biases decisions more strongly than the first one, contrary to predictions of one-stage models and intuition. We present a two-stage model where sensory information is integrated and buffered before it is fed into a drift diffusion process. The model is tested in a series of psychophysical experiments and explains both accuracy and reaction time distributions. © 2012 Rüter et al.

200-V lateral superjunction LIGBT on partial SOI

This letter presents a novel lateral superjunction lateral insulated-gate bipolar transistor (LIGBT) in partial silicon-on-insulator (SOI) technology in 0.18-μm partial-SOI (PSOI) high-voltage (HV) process. For an n-type superjunction LIGBT, the p-layer in the superjunction drift region not only helps in achieving uniform electric field distribution but also contributes to the on-state current. The superjunction LIGBT successfully achieves a breakdown voltage (BV) of 210 V with an R dson of 765 mΩ ̇ mm 2. It exhibits half the value of specific on-state resistance R dson and three times higher saturation current (I dsat) for the same BV, compared to a comparable lateral superjunction laterally diffused metal-oxide-semiconductor fabricated in the same technology. It also performs well in higher temperature dc operation with 38.8% increase in R dson at 175°C, compared to the room temperature without any degradation in latch-up performance. To realize this device, it only requires one additional mask layer into X-FAB 0.18-μm PSOI HV process. © 2012 IEEE.

Superjunction IGBT filling the gap between SJ MOSFET and ultrafast IGBT

In this letter, we report E off-versus-V ce tradeoff curves for vertical superjunction insulated-gate bipolar transistors (SJ IGBTs), exhibiting unusual inverse slopes dE off/dV ce > 0 in a transition region between purely unipolar and strongly bipolar device behaviors. This effect is due to the action of p-pillar hole current when depleting the drift layer of SJ IGBTs during turnoff and the impact of current gain on the transconductance. Such SJ IGBTs surpass by a very significant margin their superjunction MOSFET counterparts in terms of power-handling capability and on-state and turnoff losses, all at the same time. © 2012 IEEE.

Modelling and Fragility Analysis of Non-Ductile Reinforced Concrete Buildings in Low-to-Moderate Seismic Zones

Reinforced concrete buildings in low-to-moderate seismic zones are often designed only for gravity loads in accordance with the non-seismic detailing provisions. Deficient detailing of columns and beam-column joints can lead to unpredictable brittle failures even under moderate earthquakes. Therefore, a reliable estimate of structural response is required for the seismic evaluation of these structures. For this purpose, analytical models for both interior and exterior slab-beam-column subassemblages and for a 1/3 scale model frame were implemented into the nonlinear finite element platform OpenSees. Comparison between the analytical results and experimental data available in the literature is carried out using nonlinear pushover analyses and nonlinear time history analysis for the subassemblages and the model frame, respectively. Furthermore, the seismic fragility assessment of reinforced concrete buildings is performed on a set of non-ductile frames using nonlinear time history analyses. The fragility curves, which are developed for various damage states for the maximum interstory drift ratio are characterized in terms of peak ground acceleration and spectral acceleration using a suite of ground motions representative of the seismic hazard in the region.

The lateral superjunction PSOI LIGBT and LDMOSFET

This paper presents a comparison between the superjunction LIGBT and the LDMOSFET in partial silicon-on-insulator (PSOI) technology in 0.18μm PSOI HV process. The superjunction drift region helps in achieving uniform electric field distribution in both structures but also contributes to the on-state current in the LIGBT. The superjunction LIGBT successfully achieves breakdown voltage (BV) of 210V with Rdson of 765mΩ.mm2. It exhibits reduced specific on-state resistance Rdson and higher saturation current (Idsat) for the same BV compared to a compatible lateral superjunction LDMOS in the same technology. © 2012 IEEE.

Uncooled dense WDM

An uncooled dense WDM system is described using standard WDM components and receiver signal processing, with a different number of receivers to transmitters, to allow wide temperature drift of the transmitter lasers. © 2012 OSA.

Modeling bipolar power semiconductor devices gachovska

This book presents physics-based models of bipolar power semiconductor devices and their implementation in MATLAB and Simulink. The devices are subdivided into different regions, and the operation in each region, along with the interactions at the interfaces which are analyzed using basic semiconductor physics equations that govern their behavior. The Fourier series solution is used to solve the ambipolar diffusion equation in the lightly doped drift region of the devices. In addition to the external electrical characteristics, internal physical and electrical information, such as the junction voltages and the carrier distribution in different regions of the device, can be obtained using the models. Table of Contents: Introduction to Power Semiconductor Device Modeling/Physics of Power Semiconductor Devices/Modeling of a Power Diode and IGBT/IGBT Under an Inductive Load-Switching Condition in Simulink/Parameter Extraction. © 2013 by Morgan & Claypool.

Particulate matter emission during start-up and transient operation of a spark-ignition engine

In order to understand why emissions of Particulate Matter (PM) from Spark-Ignition (SI) automobiles peak during periods of transient operation such as rapid accelerations, a study of controlled, repeatable transients was performed. Time-resolved engine-out PM emissions from a modern four-cylinder engine during transient load and air/fuel ratio operation were examined, and the results could be fit in most cases to a first order time response. The time constants for the transient response are similar to those measured for changes in intake valve temperature, reflecting the strong dependence of PM emissions on the amount of liquid fuel in the combustion chamber. In only one unrepeatable case did the time response differ from a first order function: showing an overshoot in PM emissions during transition from the initial to the final steady state PM emission level. PM emissions during controlled, motored start-up experiments show a peak at start-up followed by a period during which emissions are either relatively constant or drift somewhat. When the fuel injection and ignition are shut off, PM emissions also peak briefly, but rapidly decay to low levels. Qualitative implications on the study and modeling of PM emissions during transient engine operation are discussed. Copyright © 1999 Society of Automotive Engineers, Inc.

Some remarks about periodic feedback stabilization

Periodic feedback stabilization is a very natural solution to overcome the topological obstructions which may occur when one tries to asymptotically (locally) stabilize a (locally) controllable nonlinear system around an equilibrium point. The object of this paper is to give a simple geometric interpretation of this fact, to show that one obtains a weakened form of those obstructions when periodic feedback is used, and to illustrate the success of periodic feedback stabilization on a representative system which contains a drift.

100 Gb/s uncooled WDM system using conventional WDM components and advanced receiver signal processing

We demonstrate an uncooled WDM system using standard WDM components and receiver signal processing, with a different number of receivers to transmitters, to allow wide temperature drift of the transmitter lasers. A 100 Gb/s 8-wavelength demonstrator has been developed, which proves the feasibility of the approach over 25 km of SMF. © 2012 OSA.

Improved switching uniformity and low-voltage operation in TaO x-based RRAM using Ge reactive layer

Significant improvements in the spatial and temporal uniformities of device switching parameters are successfully demonstrated in Ge/TaOx bilayer-based resistive switching devices, as compared with non-Ge devices. In addition, the reported Ge/TaOx devices also show significant reductions in the operation voltages. Influence of the Ge layer on the resistive switching of TaOx-based resistive random access memory is investigated by X-ray spectroscopy and the theory of Gibbs free energy. Higher uniformity is attributed to the confinement of the filamentary switching process. The presence of a larger number of interface traps, which will create a beneficial electric field to facilitate the drift of oxygen vacancies, is believed to be responsible for the lower operation voltages in the Ge/TaO x devices. © 1980-2012 IEEE.

Impact of donor traps on the 2DEG and electrical behavior of AlGaN/GaN MISFETs

As an important step in understanding trap-related mechanisms in AlGaN/GaN transistors, the physical properties of surface states have been analyzed through the study of the transfer characteristics of a MISFET. This letter focused initially on the relationship between donor parameters (concentration and energy level) and electron density in the channel in AlGaN/GaN heterostructures. This analysis was then correlated to dc and pulsed measurements of the transfer characteristics of a MISFET, where the gate bias was found to modulate either the channel density or the donor states. Traps-free and traps-frozen TCAD simulations were performed on an equivalent device to capture the donor behavior. A donor concentration of 1.14× 1013 ∼ cm-2 with an energy level located 0.2 eV below the conduction band edge gave the best fit to measurements. With the approach described here, we were able to analyze the region of the MISFET that corresponds to the drift region of a conventional HEMT. © 1980-2012 IEEE.