arXiv subjects

Yang Jiang

Publications and source records attributed to Yang Jiang.

43 records · Page 3Linked to original sources

arXiv2020

ivmodel: An R Package for Inference and Sensitivity Analysis of Instrumental Variables Models with One Endogenous Variable

We present a comprehensive R software ivmodel for analyzing instrumental variables with one endogenous variable. The package implements a general class of estimators called k- class estimators and two confidence intervals that are fully robust to weak instruments. The package also provides power formulas for various test statistics in instrumental variables. Finally, the package contains methods for sensitivity analysis to examine the sensitivity of the inference to instrumental variables assumptions. We demonstrate the software on the data set from Card (1995), looking at the causal effect of levels of education on log earnings where the instrument is proximity to a four-year college.

stat.AP↗

arXiv2019

Neural Architecture Refinement: A Practical Way for Avoiding Overfitting in NAS

Neural architecture search (NAS) is proposed to automate the architecture design process and attracts overwhelming interest from both academia and industry. However, it is confronted with overfitting issue due to the high-dimensional search space composed by operator selection and skip connection of each layer. This paper explores the architecture overfitting issue in depth based on the reinforcement learning-based NAS framework. We show that the policy gradient method has deep correlations with the cross entropy minimization. Based on this correlation, we further demonstrate that, though the reward of NAS is sparse, the policy gradient method implicitly assign the reward to all operations and skip connections based on the sampling frequency. However, due to the inaccurate reward estimation, curse of dimensionality problem and the hierachical structure of neural networks, reward charateristics for operators and skip connections have intrinsic differences, the assigned rewards for the skip connections are extremely noisy and inaccurate. To alleviate this problem, we propose a neural architecture refinement approach that working with an initial state-of-the-art network structure and only refining its operators. Extensive experiments have demonstrated that the proposed method can achieve fascinated results, including classification, face recognition etc.

cs.LG↗

arXiv2019

Gate Leakage Suppression and Breakdown Voltage Enhancement in p-GaN HEMTs using Metal/Graphene Gates

In this work, single-layer intrinsic and fluorinated graphene were investigated as gate insertion layers in normally-OFF p-GaN gate HEMTs, which wraps around the bottom of the gate forming Ti/graphene/p-GaN at the bottom and Ti/graphene/ SiNx on the two sides. Compared to the Au/Ti/p-GaN HEMTs without graphene, the insertion of graphene can increase the ION/IOFF ratios by a factor of 50, increase the VTH by 0.30 V and reduce the off-state gate leakage by 50 times. Additionally, this novel gate structure has better thermal stability. After thermal annealing at 350 °C, gate breakdown voltage holds at 12.1 V, which is first reported for Schottky gate p-GaN HEMTs. This is considered to be a result of the 0.24 eV increase in Schottky barrier height and the better quality of the Ti/graphene/p-GaN and Ti/graphene/SiNx interfaces. This approach is very effective in improving the Ion/Iff ratio and gate BV of normally-OFF GaN HEMTs.

physics.app-ph↗

arXiv2019

Achieving sub-1 Ohm-mm Non-Recess S/D Contact Resistance in GaN HEMTs Utilizing Simple CMOS Compatible La/Ti/Al/Ti Metal Contacts

In this paper, we report the use of lanthanum (La) in S/D contacts of GaN HEMTs, achieving 0.97 Ohm-mm contact resistance without S/D recess. The HEMTs show well-behaved electrical characteristics and satisfactory reliability. Our studies show that La, a CMOS compatible metal, is promising to lower GaN HEMT S/D contact resistance. La's low work function (3.5 eV) is beneficial for reducing the barrier between the metals and GaN. The Ohmic contact formation mechanism involved was shown to be different from conventional Ti/Al films. Spherical-shaped high-La regions formed near the surface during annealing. La diffuses into the AlGaN layer, and the overlap of La and Al peaks is significantly increased compared with that before annealing.

physics.app-ph↗

arXiv2017

Real-time brain machine interaction via social robot gesture control

Brain-Machine Interaction (BMI) system motivates interesting and promising results in forward/feedback control consistent with human intention. It holds great promise for advancements in patient care and applications to neurorehabilitation. Here, we propose a novel neurofeedback-based BCI robotic platform using a personalized social robot in order to assist patients having cognitive deficits through bilateral rehabilitation and mental training. For initial testing of the platform, electroencephalography (EEG) brainwaves of a human user were collected in real time during tasks of imaginary movements. First, the brainwaves associated with imagined body kinematics parameters were decoded to control a cursor on a computer screen in training protocol. Then, the experienced subject was able to interact with a social robot via our real-time BMI robotic platform. Corresponding to subject's imagery performance, he/she received specific gesture movements and eye color changes as neural-based feedback from the robot. This hands-free neurofeedback interaction not only can be used for mind control of a social robot's movements, but also sets the stage for application to enhancing and recovering mental abilities such as attention via training in humans by providing real-time neurofeedback from a social robot.

cs.HC↗

arXiv2017

An Effective Training Method For Deep Convolutional Neural Network

In this paper, we propose the nonlinearity generation method to speed up and stabilize the training of deep convolutional neural networks. The proposed method modifies a family of activation functions as nonlinearity generators (NGs). NGs make the activation functions linear symmetric for their inputs to lower model capacity, and automatically introduce nonlinearity to enhance the capacity of the model during training. The proposed method can be considered an unusual form of regularization: the model parameters are obtained by training a relatively low-capacity model, that is relatively easy to optimize at the beginning, with only a few iterations, and these parameters are reused for the initialization of a higher-capacity model. We derive the upper and lower bounds of variance of the weight variation, and show that the initial symmetric structure of NGs helps stabilize training. We evaluate the proposed method on different frameworks of convolutional neural networks over two object recognition benchmark tasks (CIFAR-10 and CIFAR-100). Experimental results showed that the proposed method allows us to (1) speed up the convergence of training, (2) allow for less careful weight initialization, (3) improve or at least maintain the performance of the model at negligible extra computational cost, and (4) easily train a very deep model.

cs.LG↗

arXiv2017

Brain Computer Interface for Gesture Control of a Social Robot: an Offline Study

Brain computer interface (BCI) provides promising applications in neuroprosthesis and neurorehabilitation by controlling computers and robotic devices based on the patient's intentions. Here, we have developed a novel BCI platform that controls a personalized social robot using noninvasively acquired brain signals. Scalp electroencephalogram (EEG) signals are collected from a user in real-time during tasks of imaginary movements. The imagined body kinematics are decoded using a regression model to calculate the user-intended velocity. Then, the decoded kinematic information is mapped to control the gestures of a social robot. The platform here may be utilized as a human-robot-interaction framework by combining with neurofeedback mechanisms to enhance the cognitive capability of persons with dementia.

cs.RO↗