arXiv subjects

Xun Liu

Publications and source records attributed to Xun Liu.

60 records · Page 4Linked to original sources

arXiv2018

Phased Microphone Array for Sound Source Localization with Deep Learning

To phased microphone array for sound source localization, algorithm with both high computational efficiency and high precision is a persistent pursuit. In this paper convolutional neural network (CNN) a kind of deep learning is preliminarily applied as a new algorithm. At high frequency CNN can reconstruct the sound localizations with excellent spatial resolution as good as DAMAS, within a very short time as short as conventional beamforming. This exciting result means that CNN perfectly finds source distribution directly from cross-spectral matrix without given propagation function in advance, and thus CNN deserves to be further explored as a new algorithm.

eess.AS↗

arXiv2018

Expectation Learning for Adaptive Crossmodal Stimuli Association

The human brain is able to learn, generalize, and predict crossmodal stimuli. Learning by expectation fine-tunes crossmodal processing at different levels, thus enhancing our power of generalization and adaptation in highly dynamic environments. In this paper, we propose a deep neural architecture trained by using expectation learning accounting for unsupervised learning tasks. Our learning model exhibits a self-adaptable behavior, setting the first steps towards the development of deep learning architectures for crossmodal stimuli association.

cs.LG↗

arXiv2017

PBODL : Parallel Bayesian Online Deep Learning for Click-Through Rate Prediction in Tencent Advertising System

We describe a parallel bayesian online deep learning framework (PBODL) for click-through rate (CTR) prediction within today's Tencent advertising system, which provides quick and accurate learning of user preferences. We first explain the framework with a deep probit regression model, which is trained with probabilistic back-propagation in the mode of assumed Gaussian density filtering. Then we extend the model family to a variety of bayesian online models with increasing feature embedding capabilities, such as Sparse-MLP, FM-MLP and FFM-MLP. Finally, we implement a parallel training system based on a stream computing infrastructure and parameter servers. Experiments with public available datasets and Tencent industrial datasets show that models within our framework perform better than several common online models, such as AdPredictor, FTRL-Proximal and MatchBox. Online A/B test within Tencent advertising system further proves that our framework could achieve CTR and CPM lift by learning more quickly and accurately.

cs.LG↗

arXiv2016

Improving the Efficiency of DAMAS for Sound Source Localization via Wavelet Compression Computational Grid

Phased microphone arrays are used widely in the applications for acoustic source localization. Deconvolution approaches such as DAMAS successfully overcome the spatial resolution limit of the conventional delay-and-sum (DAS) beamforming method. However deconvolution approaches require high computational effort compared to conventional DAS beamforming method. This paper presents a novel method that serves to improve the efficiency of DAMAS via wavelet compression computational grid rather than via optimizing DAMAS algorithm. In this method, the efficiency of DAMAS increases with compression ratio. This method can thus save lots of run time in industrial applications for sound source localization, particularly when sound sources are just located in a small extent compared with scanning plane and a band of angular frequency needs to be calculated. In addition, this method largely retains the spatial resolution of DAMAS on original computational grid, although with a minor deficiency that the occurrence probability of aliasing increasing slightly for complicated sound source.

cs.SD↗

arXiv2012

Plasmoid ejection and secondary current sheet generation from magnetic reconnection in laser-plasma interaction

Reconnection of the self-generated magnetic fields in laser-plasma interaction was first investigated experimentally by Nilson {\it et al.} [Phys. Rev. Lett. 97, 255001 (2006)] by shining two laser pulses a distance apart on a solid target layer. An elongated current sheet (CS) was observed in the plasma between the two laser spots. In order to more closely model magnetotail reconnection, here two side-by-side thin target layers, instead of a single one, are used. It is found that at one end of the elongated CS a fan-like electron outflow region including three well-collimated electron jets appears. The ($>1$ MeV) tail of the jet energy distribution exhibits a power-law scaling. The enhanced electron acceleration is attributed to the intense inductive electric field in the narrow electron dominated reconnection region, as well as additional acceleration as they are trapped inside the rapidly moving plasmoid formed in and ejected from the CS. The ejection also induces a secondary CS.

physics.plasm-ph↗

arXiv2007

RIP: An Efficient Hybrid Repeater Insertion Scheme for Low Power

This paper presents a novel repeater insertion algorithm for interconnect power minimization. The novelty of our approach is in the judicious integration of an analytical solver and a dynamic programming based method. Specifically, the analytical solver chooses a concise repeater library and a small set of repeater location candidates such that the dynamic programming algorithm can be performed fast with little degradation of the solution quality. In comparison with previously reported repeater insertion schemes, within comparable runtimes, our approach achieves up to 37% higher power savings. Moreover, for the same design quality, our scheme attains a speedup of two orders of magnitude.

cs.OH↗