Extending the task of diarization to speaker attribution

In this paper we extend the concept of speaker annotation within a single-recording, or speaker diarization, to a collection wide approach we call speaker attribution. Accordingly, speaker attribution is the task of clustering expectantly homogenous intersession clusters obtained using diarization according to common cross-recording identities. The result of attribution is a collection of spoken audio across multiple recordings attributed to speaker identities. In this paper, an attribution system is proposed using mean-only MAP adaptation of a combined-gender UBM to model clusters from a perfect diarization system, as well as a JFA-based system with session variability compensation. The normalized cross-likelihood ratio is calculated for each pair of clusters to construct an attribution matrix and the complete linkage algorithm is employed to conduct clustering of the inter-session clusters. A matched cluster purity and coverage of 87.1% was obtained on the NIST 2008 SRE corpus.

Automated cross-lingual link discovery in Wikipedia

At NTCIR-9, we participated in the cross-lingual link discovery (Crosslink) task. In this paper we describe our approaches to discovering Chinese, Japanese, and Korean (CJK) cross-lingual links for English documents in Wikipedia. Our experimental results show that a link mining approach that mines the existing link structure for anchor probabilities and relies on the “translation” using cross-lingual document name triangulation performs very well. The evaluation shows encouraging results for our system.

Foot pain and functional limitation in healthy adults with hallux valgus : a cross-sectional study

Background Hallux valgus (HV) is a very common deformity of the first metatarsophalangeal joint that often requires surgical correction. However, the association between structural HV deformity and related foot pain and disability is unclear. Furthermore, no previous studies have investigated concerns about appearance and difficulty with footwear in a population with HV not seeking surgical correction. The aim of this cross-sectional study was to investigate foot pain, functional limitation, concern about appearance and difficulty with footwear in otherwise healthy adults with HV compared to controls. Methods Thirty volunteers with HV (radiographic HV angle >15 degrees) and 30 matched controls were recruited for this study (50 women, 10 men; mean age 44.4 years, range 20 to 76 years). Differences between groups were examined for self-reported foot pain and disability, satisfaction with appearance, footwear difficulty, and pressure-pain threshold at the first metatarsophalangeal joint. Functional measures included balance tests, walking performance, and hallux muscle strength (abduction and plantarflexion). Mean differences (MD) and 95% confidence intervals (CI) were calculated. Results All self-report measures showed that HV was associated with higher levels of foot pain and disability and significant concerns about appearance and footwear (p < 0.001). Lower pressure-pain threshold was measured at the medial first metatarsophalangeal joint in participants with HV (MD = -133.3 kPa, CI: -251.5 to -15.1). Participants with HV also showed reduced hallux plantarflexion strength (MD = -37.1 N, CI: -55.4 to -18.8) and abduction strength (MD = -9.8 N, CI: -15.6 to -4.0), and increased mediolateral sway when standing with both feet with eyes closed (MD = 0.34 cm, CI: 0.04 to 0.63). Conclusions These findings show that HV negatively impacts on self-reported foot pain and function, and concerns about foot appearance and footwear in otherwise healthy adults. There was also evidence of impaired hallux muscle strength and increased postural sway in HV subjects compared to controls, although general physical functioning and participation in physical activity were not adversely affected.

Cross layer optimization of VANET routing with multi-objective decision making

Vehicular Ad-hoc Networks (VANET) have different characteristics compared to other mobile ad-hoc networks. The dynamic nature of the vehicles which act as routers and clients are connected with unreliable radio links and Routing becomes a complex problem. First we propose CO-GPSR (Cooperative GPSR), an extension of the traditional GPSR (Greedy Perimeter Stateless Routing) which uses relay nodes which exploit radio path diversity in a vehicular network to increase routing performance. Next we formulate a Multi-objective decision making problem to select optimum packet relaying nodes to increase the routing performance further. We use cross layer information for the optimization process. We evaluate the routing performance more comprehensively using realistic vehicular traces and a Nakagami fading propagation model optimized for highway scenarios in VANETs. Our results show that when Multi-objective decision making is used for cross layer optimization of routing a 70% performance increment can be obtained for low vehicle densities on average, which is a two fold increase compared to the single criteria maximization approach.

The backfilled GEI : a cross-capture modality gait feature for frontal and side-view gait recognition

In this paper, we propose a novel direction for gait recognition research by proposing a new capture-modality independent, appearance-based feature which we call the Back-filled Gait Energy Image (BGEI). It can can be constructed from both frontal depth images, as well as the more commonly used side-view silhouettes, allowing the feature to be applied across these two differing capturing systems using the same enrolled database. To evaluate this new feature, a frontally captured depth-based gait dataset was created containing 37 unique subjects, a subset of which also contained sequences captured from the side. The results demonstrate that the BGEI can effectively be used to identify subjects through their gait across these two differing input devices, achieving rank-1 match rate of 100%, in our experiments. We also compare the BGEI against the GEI and GEV in their respective domains, using the CASIA dataset and our depth dataset, showing that it compares favourably against them. The experiments conducted were performed using a sparse representation based classifier with a locally discriminating input feature space, which show significant improvement in performance over other classifiers used in gait recognition literature, achieving state of the art results with the GEI on the CASIA dataset.

Cross-Lingual knowledge discovery : Chinese-to-English article linking in Wikipedia

In this paper we examine automated Chinese to English link discovery in Wikipedia and the effects of Chinese segmentation and Chinese to English translation on the hyperlink recommendation. Our experimental results show that the implemented link discovery framework can effectively recommend Chinese-to-English cross-lingual links. The techniques described here can assist bi-lingual users where a particular topic is not covered in Chinese, is not equally covered in both languages, or is biased in one language; as well as for language learning.

Cross-layer design and optimization for wireless networked control systems

Wireless networked control systems (WNCSs) have been widely used in the areas of manufacturing and industrial processing over the last few years. They provide real-time control with a unique characteristic: periodic traffic. These systems have a time-critical requirement. Due to current wireless mechanisms, the WNCS performance suffers from long time-varying delays, packet dropout, and inefficient channel utilization. Current wirelessly networked applications like WNCSs are designed upon the layered architecture basis. The features of this layered architecture constrain the performance of these demanding applications. Numerous efforts have attempted to use cross-layer design (CLD) approaches to improve the performance of various networked applications. However, the existing research rarely considers large-scale networks and congestion network conditions in WNCSs. In addition, there is a lack of discussions on how to apply CLD approaches in WNCSs. This thesis proposes a cross-layer design methodology to address the issues of periodic traffic timeliness, as well as to promote the efficiency of channel utilization in WNCSs. The design of the proposed CLD is highlighted by the measurement of the underlying network condition, the classification of the network state, and the adjustment of sampling period between sensors and controllers. This period adjustment is able to maintain the minimally allowable sampling period, and also maximize the control performance. Extensive simulations are conducted using the network simulator NS-2 to evaluate the performance of the proposed CLD. The comparative studies involve two aspects of communications, with and without using the proposed CLD, respectively. The results show that the proposed CLD is capable of fulfilling the timeliness requirement under congested network conditions, and is also able to improve the channel utilization efficiency and the proportion of effective data in WNCSs.

An evaluation framework for cross-lingual link discovery

Cross-Lingual Link Discovery (CLLD) is a new problem in Information Retrieval. The aim is to automatically identify meaningful and relevant hypertext links between documents in different languages. This is particularly helpful in knowledge discovery if a multi-lingual knowledge base is sparse in one language or another, or the topical coverage in each language is different; such is the case with Wikipedia. Techniques for identifying new and topically relevant cross-lingual links are a current topic of interest at NTCIR where the CrossLink task has been running since the 2011 NTCIR-9. This paper presents the evaluation framework for benchmarking algorithms for cross-lingual link discovery evaluated in the context of NTCIR-9. This framework includes topics, document collections, assessments, metrics, and a toolkit for pooling, assessment, and evaluation. The assessments are further divided into two separate sets: manual assessments performed by human assessors; and automatic assessments based on links extracted from Wikipedia itself. Using this framework we show that manual assessment is more robust than automatic assessment in the context of cross-lingual link discovery.

Very sensitive fiber Bragg grating accelerometer using transverse forces with an easy over-range protection and low cross axial sensitivity

The first fiber Bragg grating (FBG) accelerometer using direct transverse forces is demonstrated by fixing the FBG by its two ends and placing a transversely moving inertial object at its middle. It is very sensitive because a lightly stretched FBG is more sensitive to transverse forces than axial forces. Its resonant frequency and static sensitivity are analyzed by the classic spring-mass theory, assuming the axial force changes little. The experiments show that the theory can be modified for cases where the assumption does not hold. The resonant frequency can be modified by a linear relationship experimentally achieved, and the static sensitivity by an alternative method proposed. The principles of the over-range protection and low cross axial sensitivity are achieved by limiting the movement of the FBG and were validated experimentally. The sensitivities 1.333 and 0.634 nm/g were experimentally achieved by 5.29 and 2.83 gram inertial objects at 10 Hz from 0.1 to 0.4 g (g = 9.8 m/s 2), respectively, and their resonant frequencies were around 25 Hz. Their theoretical static sensitivities and resonant frequencies found by the modifications are 1.188 nm/g and 26.81 Hz for the 5.29 gram one and 0.784 nm/g and 29.04 Hz for the 2.83 gram one, respectively.

Cross-sectional study of a microsatellite marker in the low densitity lipoprotein receptor gene in obese normotensives

1. The low density lipoprotein receptor is an important regulator of serum cholesterol which may have implications for the development of both hypertension and obesity. In this study, genotypes for a low density lipoprotein receptor gene (LDLR) dinucleotide polymorphism were determined in both lean and obese normotensive populations. 2. In previous cross-sectional association studies an ApaLI and a HincII polymorphism for LDLR were shown to be associated with obesity in essential hypertensives. However, these polymorphisms did not show an association with obesity in normotensives. 3. In contrast, this study reports that preliminary results for an LDLR microsatellite marker, located more towards the 3' end of the gene, show a significant association with obesity in the normotensive population studied. These results indicate that LDLR could play an important role in the development of obesity, which might be independent of hypertension.

A linkage and cross-sectional study of hypertension and obesity using a poly(A) Alu-repeat polymorphism at the glucagon receptor gene locus (17q25)

1. Previous glucagon receptor gene (GCGR) studies have shown a Gly40Ser mutation to be more prevalent in essential hypertension and to affect glucagon binding affinity to its receptor. An Alu-repeat poly(A) polymorphism colocalized to GCGR was used in the present study to test for association and linkage in hypertension as well as association in obesity development. 2. Using a cross-sectional approach, 85 hypertensives and 95 normotensives were genotyped using polymerase chain reaction primers flanking the Alu-repeat. Both hypertensive and normotensive populations were subdivided into lean and obese categories based on body mass index (BMI) to determine involvement of this variant in obesity. For the linkage study, 89 Australian Caucasian hypertension affected sibships (174 sibpairs) were genotyped and the results were analysed using GENE-HUNTER, Mapmaker Sibs, ERPA and SPLINK (all freely available from http://linlkage.rockefeller. edu/soft/list.html). 3. Cross-sectional results for both hypertension and obesity were analysed using Chi-squared and Monte Carlo analyses. Results did not show an association of this variant with either hypertension (χ2 = 6.9, P = 0.14; Monte Carlo χ2 = 7.0, P = 0.11; n = 5000) or obesity (χ2 = 3.3, P = 0.35; Monte Carlo χ2 = 3.26, P = 0.34; n = 5000). In addition, results from the linkage study using hypertensive sib-pairs did not indicate linkage of the poly(A) repent with hypertension. Hence, results did not indicate a role far the Alu-repeat in either hypertension or obesity. However, as the heterozygosity of this poly(A) repeat is low (35%), a larger number of hypertensive sib-pairs may be required to draw definitive conclusions.

Unusual cross-ring S(N)2 reactions of M-H (-) ions of methoxyacetanilides

Deprotonated o, m-, and p-methoxyacetanilide show pronounced peaks in their collision-induced tandem mass spectra (MS/MS) produced by losses of the elements of C2H6. It is proposed that this reaction is a 'cross-ring' internal S(N)2 reaction involving an incipient methyl anion. For example, p-CH3O-C6H4-N--CO-CH3--> [(p.CH3O-C6H4-N=C=O)CH3-]--> O---C6H4-N=C=O+C2H6.

Unusual cross-ring SN2 reactions of [MH]− ions of methoxyacetanilides

Deprotonated o, m-, and p-methoxyacetanilide show pronounced peaks in their collision-induced tandem mass spectra (MS/MS) produced by losses of the elements of C2H6. It is proposed that this reaction is a 'cross-ring' internal S(N)2 reaction involving an incipient methyl anion. For example, p-CH3O-C6H4-N--CO-CH3--> [(p.CH3O-C6H4-N=C=O)CH3-]--> O---C6H4-N=C=O+C2H6.

Phagocytosis of cross-linked gelatin matrix by human breast carcinoma cells correlates with their invasive capacity

During invasion and metastasis, cancer cells interact closely with the extracellular matrix molecules by attachment, degradation, and migration. We demonstrated previously the local degradation of fluorescently labeled gelatin matrix by cancer cells at invasive membrane protrusions, called invadopodia. Using the newly developed quantitative fluorescence-activated cell sorting-phagocytosis assay and image analysis of localized degradation of fluorescently labeled matrix, we document here that degradation and site- specific removal of cross-linked gelatin matrix is correlated with the extent of phagocytosis in human breast cancer cells. A higher phagocytic capacity is generally associated with increasing invasiveness, documented in other invasion and motility assays as well. Gelatin phagocytosis is time and cell density dependent, and it is mediated by the actin cytoskeleton. Most of the intracellular gelatin is routed to actively acidified vesicles, as demonstrated by the fluorescent colocalization of gelatin with acidic vesicles, indicating the intracellular degradation of the phagocytosed matrix in lysosomes. We show here that normal intracellular routing is blocked after treatment with acidification inhibitors. In addition, the need for partial proteolytic degradation of the matrix prior to phagocytosis is demonstrated by the inhibition of gelatin phagocytosis with different serine and metalloproteinase inhibitors and its stimulation by conditioned medium containing the matrix metalloproteinases MMP-2 and MMP-9. Our results demonstrate that phagocytosis of extracellular matrix is an inherent feature of breast tumor cells that correlates with and may even directly contribute to their invasive capacity. This assay is useful for screening and evaluating potential anti-invasive agents because it is fast, reproducible, and versatile.

An iterative speaker re-diarization scheme for improving speaker-based entity extraction in multimedia archives

In this paper we present a novel scheme for improving speaker diarization by making use of repeating speakers across multiple recordings within a large corpus. We call this technique speaker re-diarization and demonstrate that it is possible to reuse the initial speaker-linked diarization outputs to boost diarization accuracy within individual recordings. We first propose and evaluate two novel re-diarization techniques. We demonstrate their complementary characteristics and fuse the two techniques to successfully conduct speaker re-diarization across the SAIVT-BNEWS corpus of Australian broadcast data. This corpus contains recurring speakers in various independent recordings that need to be linked across the dataset. We show that our speaker re-diarization approach can provide a relative improvement of 23% in diarization error rate (DER), over the original diarization results, as well as improve the estimated number of speakers and the cluster purity and coverage metrics.