Hongjie Chen

Learn More
We adopt a Dirichlet process Gaussian mixture model (DPGMM) for unsupervised acoustic modeling and represent speech frames with Gaussian posteriorgrams. The model performs unsupervised clustering on untranscribed data, and each Gaussian component can be considered as a cluster of sounds from various speakers. The model infers its model complexity (i.e. the(More)
This study was carried out to investigate the effects of resveratrol on cigarette smoke (CS)-induced lung injury. Experimental mice were administrated with 1 mg/kg or 3 mg/ kg resveratrol orally, 1 h prior to CS exposure (five cigarettes a day for 3 consecutive days). Airway inflammation and gene expression changes were assessed. CS exposure increased the(More)
Microporosity plays a key role in bioactivity and osteoinductivity of a biomaterial scaffold. A simple new approach to fabricating load-bearing porous titanium (Ti) scaffolds with uniform porous structure, highly controllable pore size and excellent biocompatibility was developed in the present study. This method was based on stack sintering of microporous(More)
In this paper we describe the system proposed by NNI (NWPUNTU-I2R) team for the QUESST task within the Mediaeval 2014 evaluation. To solve the problem, we used both dynamic time warping (DTW) and symbolic search (SS) based approaches. The DTW system performs template matching using subsequence DTW algorithm and posterior representations. The symbolic search(More)
In this paper, we propose a partial sequence matching based symbolic search (SS) method for the task of language independent query-by-example spoken term detection. One main drawback of conventional SS approach is the high miss rate for long queries. This is due to high variations in symbol representation of query and search audios, especially in language(More)
A robust and sensitive analytical method was developed for the simultaneous analysis of 21 target antimicrobials in different environmental water samples. Both single SPE and tandem SPE cartridge systems were investigated to simultaneously extract multiple classes of antimicrobials. Experimental results showed that good extraction efficiencies (84.5-105.6%)(More)
This paper describes the system developed by the NNI team for the Query-by-Example Search on Speech Task (QUESST) in the MediaEval 2015 evaluation. Our submitted system mainly used bottleneck features/stacked bottleneck features (BNF/SBNF) trained from various resources. We investigated noise robustness techniques to deal with the noisy data of this year.(More)
Cloud computing encourages application to migrate into it for economic of scale, where they rent shared resources to deliver services. Service Level Agreements(SLA) plays an important role in assisting various applications providing high-quality services to end users in cloud's complex and uncertain environments. Most of the existing work tries to support(More)
We propose a framework which ports Dirichlet Gaussian mixture model (DPGMM) based labels to deep neural network (DNN). The DNN trained using the unsupervised labels is used to extract a low-dimensional unsupervised speech representation, named as unsupervised bottleneck features (uBNFs), which capture considerable information for sound cluster(More)
As crowdsourcing has been applied to a variety of disciplines, e.g. marketing and operationalization, more and more scientists turn their sights to how the crowd innovate software engineering to produce high quality software. However, they mainly focus on the impacts brought by domain experts or experienced developers on developing and managing open source(More)