Factors influencing robustness and effectiveness of conditional random fields in active learning frameworks

Active learning approaches reduce the annotation cost required by traditional supervised approaches to reach the same effectiveness by actively selecting informative instances during the learning phase. However, effectiveness and robustness of the learnt models are influenced by a number of factors. In this paper we investigate the factors that affect the effectiveness, more specifically in terms of stability and robustness, of active learning models built using conditional random fields (CRFs) for information extraction applications. Stability, defined as a small variation of performance when small variation of the training data or a small variation of the parameters occur, is a major issue for machine learning models, but even more so in the active learning framework which aims to minimise the amount of training data required. The factors we investigate are a) the choice of incremental vs. standard active learning, b) the feature set used as a representation of the text (i.e., morphological features, syntactic features, or semantic features) and c) Gaussian prior variance as one of the important CRFs parameters. Our empirical findings show that incremental learning and the Gaussian prior variance lead to more stable and robust models across iterations. Our study also demonstrates that orthographical, morphological and contextual features as a group of basic features play an important role in learning effective models across all iterations.

Transcriptome analyses of amoebic gill disease-affected atlantic salmon (Salmo salar) tissues reveal localized host gene suppression

The transcriptome response of Atlantic salmon (Salmo salar) displaying advanced stages of amoebic gill disease (AGD) was investigated. Naïve smolt were challenged with AGD for 19 days, at which time all fish were euthanized and their severity of infection quantified through histopathological scoring. Gene expression profiles were compared between heavily infected and naïve individuals using a 17 K Atlantic salmon cDNA microarray with real-time quantitative RT-PCR (qPCR) verification. Expression profiles were examined in the gill, anterior kidney, and liver. Twenty-seven transcripts were significantly differentially expressed within the gill; 20 of these transcripts were down-regulated in the AGD-affected individuals compared with naïve individuals. In contrast, only nine transcripts were significantly differentially expressed within the anterior kidney and five within the liver. Again the majority of these transcripts were down-regulated within the diseased individuals. A down-regulation of transcripts involved in apoptosis (procathepsin L, cathepsin H precursor, and cystatin B) was observed in AGD-affected Atlantic salmon. Four transcripts encoding genes with antioxidant properties also were down-regulated in AGD-affected gill tissue according to qPCR analysis. The most up-regulated transcript within the gill was an unknown expressed sequence tag (EST) whose expression was 218-fold (± SE 66) higher within the AGD affected gill tissue. Our results suggest that Atlantic salmon experiencing advanced stages of AGD demonstrate general down-regulation of gene expression, which is most pronounced within the gill. We propose that this general gene suppression is parasite-mediated, thus allowing the parasite to withstand or ameliorate the host response. © 2008 Springer Science+Business Media, LLC.

Heat maps for aggregating bioacoustic annotations

In our large library of annotated environmental recordings of animal vocalizations, searching annotations by label can return thousands of results. We propose a heat map of aggregated annotation time and frequency bounds, maintaining the shape of the annotations as they appear on the spectrogram. This compactly displays the distribution of annotation bounds for the user's query, and allows them to easily identify unusual annotations. Key to this is allowing zero values on the map to be differentiated from areas where there are single annotations.

Detecting rare events using Kullback-Leibler divergence

One main challenge in developing a system for visual surveillance event detection is the annotation of target events in the training data. By making use of the assumption that events with security interest are often rare compared to regular behaviours, this paper presents a novel approach by using Kullback-Leibler (KL) divergence for rare event detection in a weakly supervised learning setting, where only clip-level annotation is available. It will be shown that this approach outperforms state-of-the-art methods on a popular real-world dataset, while preserving real time performance.

On the Security of Two RFID Mutual Authentication Protocols

In this paper, the security of two recent RFID mutual authentication protocols are investigated. The first protocol is a scheme proposed by Huang et al. [7] and the second one by Huang, Lin and Li [6]. We show that these two protocols have several weaknesses. In Huang et al.’s scheme, an adversary can determine the 32-bit secret password with a probability of 2−2 , and in Huang-Lin-Li scheme, a passive adversary can recognize a target tag with a success probability of 1−2−4 and an active adversary can determine all 32 bits of Access password with success probability of 2−4 . The computational complexity of these attacks is negligible.

Cryptanalysis of Tav-128 Hash Function

Many RFID protocols use cryptographic hash functions for their security. The resource constrained nature of RFID systems forces the use of light weight cryptographic algorithms. Tav-128 is one such 128-bit light weight hash function proposed by Peris-Lopez et al. for a low-cost RFID tag authentication protocol. Apart from some statistical tests for randomness by the designers themselves, Tav-128 has not undergone any other thorough security analysis. Based on these tests, the designers claimed that Tav-128 does not posses any trivial weaknesses. In this article, we carry out the first third party security analysis of Tav-128 and show that this hash function is neither collision resistant nor second preimage resistant. Firstly, we show a practical collision attack on Tav-128 having a complexity of 237 calls to the compression function and produce message pairs of arbitrary length which produce the same hash value under this hash function. We then show a second preimage attack on Tav-128 which succeeds with a complexity of 262 calls to the compression function. Finally, we study the constituent functions of Tav-128 and show that the concatenation of nonlinear functions A and B produces a 64-bit permutation from 32-bit messages. This could be a useful light weight primitive for future RFID protocols.

Practical attack on NLM-MAC scheme

The NLM stream cipher designed by Hoon Jae Lee, Sang Min Sung, Hyeong Rag Kim is a strengthened version of the LM summation generator that combines linear and non-linear feedback shift registers. In recent works, the NLM cipher has been used for message authentication in lightweight communication over wireless sensor networks and for RFID authentication protocols. The work analyses the security of the NLM stream cipher and the NLM-MAC scheme that is built on the top of the NLM cipher. We first show that the NLM cipher suffers from two major weaknesses that lead to key recovery and forgery attacks. We prove the internal state of the NLM cipher can be recovered with time complexity about nlog7×2, where the total length of internal state is 2⋅n+22⋅n+2 bits. The attack needs about n2n2 key-stream bits. We also show adversary is able to forge any MAC tag very efficiently by having only one pair (MAC tag, ciphertext). The proposed attacks are practical and break the scheme with a negligible error probability.

Using viewer’s facial expression and heart rate for sports video highlights detection

Viewer interests, evoked by video content, can potentially identify the highlights of the video. This paper explores the use of facial expressions (FE) and heart rate (HR) of viewers captured using camera and non-strapped sensor for identifying interesting video segments. The data from ten subjects with three videos showed that these signals are viewer dependent and not synchronized with the video contents. To address this issue, new algorithms are proposed to effectively combine FE and HR signals for identifying the time when viewer interest is potentially high. The results show that, compared with subjective annotation and match report highlights, ‘non-neutral’ FE and ‘relatively higher and faster’ HR is able to capture 60%-80% of goal, foul, and shot-on-goal soccer video events. FE is found to be more indicative than HR of viewer’s interests, but the fusion of these two modalities outperforms each of them.

Variation in the complex carbohydrate biosynthesis loci of Acinetobacter baumannii genomes

Extracellular polysaccharides are major immunogenic components of the bacterial cell envelope. However, little is known about their biosynthesis in the genus Acinetobacter, which includes A. baumannii, an important nosocomial pathogen. Whether Acinetobacter sp. produce a capsule or a lipopolysaccharide carrying an O antigen or both is not resolved. To explore these issues, genes involved in the synthesis of complex polysaccharides were located in 10 complete A. baumannii genome sequences, and the function of each of their products was predicted via comparison to enzymes with a known function. The absence of a gene encoding a WaaL ligase, required to link the carbohydrate polymer to the lipid A-core oligosaccharide (lipooligosaccharide) forming lipopolysaccharide, suggests that only a capsule is produced. Nine distinct arrangements of a large capsule biosynthesis locus, designated KL1 to KL9, were found in the genomes. Three forms of a second, smaller variable locus, likely to be required for synthesis of the outer core of the lipid A-core moiety, were designated OCL1 to OCL3 and also annotated. Each K locus includes genes for capsule export as well as genes for synthesis of activated sugar precursors, and for glycosyltransfer, glycan modification and oligosaccharide repeat-unit processing. The K loci all include the export genes at one end and genes for synthesis of common sugar precursors at the other, with a highly variable region that includes the remaining genes in between. Five different capsule loci, KL2, KL6, KL7, KL8 and KL9 were detected in multiply antibiotic resistant isolates belonging to global clone 2, and two other loci, KL1 and KL4, in global clone 1. This indicates that this region is being substituted repeatedly in multiply antibiotic resistant isolates from these clones.

The O-specific polysaccharide structure and biosynthetic gene cluster of Yersinia pseudotuberculosis serotype O:11

In the Yersinia pseudotuberculosis serotyping scheme, 21 serotypes are present originating from about 30 different O-factors distributed within the species. With regard to the chemical structures of lipopolysaccharides (LPSs) and the genetic basis of their biosynthesis, a number, but not all, of Y. pseudotuberculosis strains representing different serotypes have been investigated. In order to present an overall picture of the relationship between genetics and structures, we have been working on the genetics and structures of various Y. pseudotuberculosis O-specific polysaccharides (OPSs). Here, we present a structural and genetic analysis of the Y. pseudotuberculosis serotype O:11 OPS. Our results showed that this OPS structure has the same backbone as that of Y. pseudotuberculosis O:1b, but with a 6d-l-Altf side-branch instead of Parf. The 3′ end of the gene cluster is the same as that for O:1b and has the genes for synthesis of the backbone and for processing the completed repeat unit. The 5′ end has genes for synthesis of 6d-l-Altf and its transfer to the repeating unit backbone. The pathway for the synthesis of the 6d-l-Altf appears to be different from that for 6d-l-Altp in Y. enterocolitica O:3. The chemical structure of the O:11 repeating unit is [Figure]

The genetics and structure of the O-specific polysaccharide of Yersinia pseudotuberculosis serotype O:10 and its relationship with Escherichia coli O111 and Salmonella enterica O35

The O-specific polysaccharide (OPS) is a variable constituent of the lipopolysaccharide of Gram-negative bacteria. The polymorphic nature of OPSs within a species is usually first defined serologically, and the current serotyping scheme for Yersinia pseudotuberculosis consists of 21 O serotypes of which 15 have been characterized genetically and structurally. Here, we present the structure and DNA sequence of Y. pseudotuberculosis O:10 OPS. The O unit consists of one residue each of d-galactopyranose, N-acetyl-d-galactosamine (2-amino-2-deoxy-d-galactopyranose) and d-glucopyranose in the backbone, with two colitose (3,6-dideoxy-l-xylo-hexopyranose) side-branch residues. This structure is very similar to that shared by Escherichia coli O111 and Salmonella enterica O35. The gene cluster sequences of these serotypes, however, have only low levels of similarity to that of Y. pseudotuberculosis O:10, although there is significant conservation of gene order. Within Y. pseudotuberculosis, the O10 structure is most closely related to the O:6 and O:7 structures.

Genetic characterisation and structural analysis of the O-specific polysaccharide of Yersinia pseudotuberculosis serotype O:1c

Many, but not all, of the current 21 serotypes of Yersinia pseudotuberculosis have been investigated with regard to the chemical structures of their O-specific polysaccharide (OPS) and the genetic basis of their biosynthesis. Completion of the genetics and structures of the remaining serotypes will enhance our understanding of the emerging relationship between genetics and structures within this species. Here, we present a structural and genetic analysis of the Y. pseudotuberculosis serotype O:1c OPS. Our results showed that this OPS has the same backbone as Y. pseudotuberculosis O:2b, but with a 3,6-dideoxy-D-ribo-hexofuranose (paratofuranose, Parf) side-branch instead of a 3,6-dideoxy-D-xylo-hexopyranose (abequopyranose, Abep). The 3'-end of the gene cluster is the same as for O:2b and has the genes for synthesis of the backbone and for processing the completed repeat unit. The 5'-end of the cluster consists of the same genes as O:1b for synthesis of Parf and a related gene for its transfer to the repeating unit backbone.

The O-specific polysaccharide structure and gene cluster of serotype O:12 of the Yersinia pseudotuberculosis complex, and the identification of a novel L-quinovose biosynthesis gene

A major virulence factor for Yersinia pseudotuberculosis is lipopolysaccharide, including O-polysaccharide (OPS). Currently, the OPS based serotyping scheme for Y. pseudotuberculosis includes 21 known O-serotypes, with genetic and structural data available for 17 of them. The completion of the OPS structures and genetics of this species will enable the visualization of relationships between O-serotypes and allow for analysis of the evolutionary processes within the species that give rise to new serotypes. Here we present the OPS structure and gene cluster of serotype O:12, thus adding one more to the set of completed serotypes, and show that this serotype is present in both Y. pseudotuberculosis and the newly identified Y. similis species. The O:12 structure is shown to include two rare sugars: 4-C[(R)-1-hydroxyethyl]-3,6-dideoxy-d-xylo-hexose (d-yersiniose) and 6-deoxy-l-glucopyranose (l-quinovose). We have identified a novel putative guanine diphosphate (GDP)-l-fucose 4-epimerase gene and propose a pathway for the synthesis of GDP-l-quinovose, which extends the known GDP-l-fucose pathway.