Azimuth Estimation for Sectorized Base Station With Improved Soft-Margin Classification

Soft-Margin Softmax for Deep Classification

In deep classification, the softmax loss (Softmax) is arguably one of the most commonly used components to train deep convolutional neural networks (CNNs). However, such a widely used loss is limited due to its lack of encouraging the discriminability of features. Recently, the large-margin softmax loss (L-Softmax [14]) is proposed to explicitly enhance the feature discrimination, with hard mar...

Soft margin estimation for automatic speech recognition

Soft margin estimation with various separation levels for LVCSR

We continue our previous work on soft margin estimation (SME) to large vocabulary continuous speech recognition (LVCSR) in two new aspects. The first is to formulate SME with different unit separation. SME methods focusing on string-, word-, and phonelevel separation are defined. The second is to compare SME with all the popular conventional discriminative training (DT) methods, including maxim...

An Improved Method of Pose Estimation for Lighthouse Base Station Extension

In 2015, HTC and Valve launched a virtual reality headset empowered with Lighthouse, the cutting-edge space positioning technology. Although Lighthouse is superior in terms of accuracy, latency and refresh rate, its algorithms do not support base station expansion, and is flawed concerning occlusion in moving targets, that is, it is unable to calculate their poses with a small set of sensors, r...

Multi-azimuth velocity estimation

It is well known that the inverse problem of estimating interval velocities from reflection data is poorly constrained in 2-D. We show that the interval velocity estimation problem in 3-D is much better constrained when the velocity function is estimated by jointly inverting the data collected along multiple offset-azimuth, even when the azimuth range is fairly limited. We extend to 3-D the lin...