Tag Archives: faah inhibitors

While T cell activity has not

While γδ T cell activity has not been well characterized in the context of BLV infection, bovine γδ T cell activity in the context of many other infections has been investigated. Cattle are considered to have a high proportion of γδ T faah inhibitors in the periphery, which is suggestive that they are of particular importance to cattle immunity (Baldwin and Telfer, 2015). Indeed, evidence suggests that γδ T cells in cattle can form memory populations in response to vaccination (Blumerman et al., 2007). In addition, WC1+ γδ T cells in cattle can produce IFNγ in response to stimulation, and this may skew immune polarization to a Th1 phenotype (Baldwin and Telfer, 2015). However, at this point it is unclear why γδ T cells from BLV+ cows have increased reactivity to stimulation in vitro, although it is possible that BLV+ cows have a higher ratio of WC1+: WC1- γδ T cells, which could explain the higher proportion of IFNγ-producing γδ T cells observed. It is also important to note that short term in vitro PBMC culture conditions can induce BLV protein expression (Frie and Coussens, 2015), so it is possible that increased reactivity of γδ T cells isolated from BLV+ cows is the result of γδ T cells that are reactive against BLV antigens, not the selected culture stimulant. However, under these culture conditions we were unable to detect BLV gp51 protein expression.
In this study, vaccination was used to determine if BLV+ cows mount a weaker immune response to stimulation. BLV+ cows tended to produce lower antigen-specific IgM and IgG2 antibody titers and displayed atypical B and T cell responses to in vitro stimulation after boost vaccination. These data support the hypothesis that BLV+ cows develop abnormal immune responses to stimulation and likely have compromised protection against other infectious diseases. The plasma antibody titers in conjunction with in vitro B and T cell activation data also warrant further investigation to understand the mechanism by which BLV infection interferes with antibody production. Further understanding of BLV’s effect on the bovine immune system is critical to ensure that the US dairy industry is equipped to assess and address the threat of BLV infection to dairy herds nationwide.

Acknowledgements
The authors gratefully acknowledge the contributions of other members of the Molecular Pathogenesis Laboratory: Dr. Jonathon Roussey, Jenna Carter, Hannah Dewald, Rachel Courville, Kristina Meier and Thilo Hamlischer. The authors also would like to thank the manager of the Michigan State University Dairy Teaching and Research Center, Robert West, as well as Kerry Nobis and the Nobis Dairy Farms. This work was supported by the United States Department of Agriculture and the National Institute of Food and Agriculture (2014-67015-21632, 2014-68004-21881, and 2016-67011-24713), Michigan AgBioResearch, the Michigan Alliance for Animal Agriculture and the Michigan Milk Producers Association.

Introduction
In swine production, weaning is a stressful period that reduces growth performance (Pluske et al., 1997), induces major perturbation of the intestinal microbiota (Castillo et al., 2007), and impairs intestinal barrier and immune functions and increases susceptibility to pathogenic bacteria such as enterotoxigenic Escherichia coli (Schroyen et al., 2013). Moreover, selection for improved sow prolificacy has resulted in a greater number of newborn piglets per litter and an increased proportion of low-birth-weight (LW) piglets within litters (Milligan et al., 2002; Quesnel et al., 2008). Therefore, such heterogeneity of birth weight induced by hyperprolificity generates distinct populations of piglets of low weight (LW) and high weight (HW) within litters and is likely to affect the access of piglets to colostrum and milk during the lactation, with consequences for future growth performance and resistance to diseases (Douglas et al., 2014). Indeed, the growth performance of LW piglets during lactation and after weaning is impaired in comparison with that of heavier piglets (Quiniou et al., 2002; Berard et al., 2008; Beaulieu et al., 2010). There is also evidence that piglet weight at birth and environmental conditions influence immune competence development (Schokker et al., 2014; Hu et al., 2015).

faah inhibitors br Introduction When a target with specified

Introduction
When a target with specified dimensions is insonified by an incident elastic/acoustic wave, in addition to reflection, refraction and mode conversion, waves may also scatter from the target into the surrounding media. Upon the interaction of the incident wave with the elastic target, surface waves are generated around the periphery of the cylinders and upon constructive interference, they build up to the resonance modes of the target. By examining the resonance frequencies and resonance modes, valuable information regarding the characteristics (both geometry and material) of the target is obtained. This procedure is called Resonance Acoustic Spectroscopy (RAS). In RAS, a broad-band acoustic wave is used to excite the resonant modes of an immersed target. If the wave frequency coincides with one of the resonance frequencies of the target, the corresponding resonance mode of the immersed body is excited. Consequently, the body starts vibrating and energy is dissipated into the surrounding medium.
Several studies have been conducted by various researches to reveal the effects of surface waves on resonance phenomenon. In these investigations, it has been shown that when the frequency of the incident wave coincides with one of the natural frequencies of the cylinder, the surface waves interfere constructively and lead to the formation of standing waves around the cylinder. This is the basis of Resonance Scattering Theory (RST) which was firstly introduced by Flax et al. [1]. Talmant et al. [2] used RST for studying the scattering of acoustic waves from thin air-filled aluminium shells. To investigate the scattering of ultrasonic waves from anisotropic cylinders, de Billy conducted a number of experiments [3]. Employing the method of normal mode expansion, Honarvar and Sinclair [4] presented an analytical solution for the problem of scattering of faah inhibitors waves from a transversely isotropic cylinder immersed in a fluid. Following the same procedure, Fan et al. [5] used the method of normal mode expansion to develop a new analytical model for scattering of ultrasonic waves from a transversely isotropic cylinder encased in an elastic matrix. Biwa et al. [6] presented a computational method for multiple scattering of transverse waves from a fiber-reinforced composite material. Jamali et al. [7] reported on the scattering of elastic waves from a cylinder made of functionally graded materials (FGM). The method was developed for two dimensional problems by employing the T-matrix method. In recent years, this field of research has attracted a great deal of interest due to new developments in the nondestructive evaluation (NDE) methods.
Lethuillier et al. [8] performed theoretical and experimental studies of acoustic wave scattering from a finite linear grating of elastic cylindrical shells. It has been observed that a resonant interaction occur at low frequencies where the shells are very close to each other. Le Bas et al. [9] showed that this phenomenon is also seen in the case of a triangular grating of immersed cylindrical shells. Robert et al. [10] considered multiple scattering by a finite number of closely-spaced cylindrical cavities (both empty and fluid-filled) embedded in an elastic matrix. They showed that, with the exception of a few narrow resonances, the single fluid-filled cavity resonances do not display the splitting phenomenon seen for immersed elastic scatterers.
Cai [11] presented a closed form solution for scattering of antiplane shear waves by a layered circular elastic cylinder embedded in a solid elastic medium. The scattering of a plane acoustic wave by an infinite penetrable or impenetrable circular cylinder, parallel with another one, of acoustically small radius, is presented by Roumeliotis et al. [12]. The method of separation of variables, in conjunction with translational addition theorems for cylindrical wave functions, was used in their study. The eigenmodes of the time-reversal operator are studied for two elastic cylinders by Minonzio et al. [13]. The effects of multiple scattering and anisotropic scattering are considered in their study. The resonance scattering by a periodic infinite array of fluid-filled cylindrical cavities in an elastic matrix is studied by Robert et al. [14]. Chen et al. [15] studied the multiple scattering of sound waves from circular cylinders embedded in an elastic medium.

br Staging Accurate determination of stage MCTs requires proper identification

Staging
Accurate determination of stage 2 MCTs requires proper identification of the draining lymph node, as well as detection of metastatic spread to this lymph node. Staging of canine cutaneous MCTs as stage 1 or 2 is clinically important, since the difference in survival time between these stages has been reported as 6.2 versus 0.8 years, respectively (Hayes et al., 2007). One study suggests that mapping of sentinel nodes may be required to accurately select the lymph node with the highest risk of metastatic disease (Worley, 2014). According to this study, 8/19 (42%) dogs had sentinel nodes with MCT metastases that differed from the lymph node in the anatomically closest location (Worley, 2014). In addition to accurately identifying the draining lymph node, proper staging is often hampered by the inability to differentiate neoplastic from non-neoplastic mast cells. Solitary mast cells are commonly observed within lymph nodes, especially with special stains, but may represent ‘inflammatory’ rather than neoplastic cells.
Through convention, most pathologists only consider large faah inhibitors of mast cells as evidence of nodal metastasis. To standardize the interpretation of nodal MCT metastasis, two classification schemes for nodal mast cell infiltrates have been recently proposed for fine needle lymph node aspirates and nodal biopsies, respectively (Krick et al., 2009; Weishaar et al., 2014). Both classifications linked the different groups to survival times. Using fine needle aspirates (FNAs), 2–3 aggregates of 2–3 mast cells are considered as possible metastases, >3 aggregates of 2–3 mast cells and/or 2–5 aggregates of >3 mast cells represent probable metastases, and certain metastasis is characterized by large numbers of mast cells, and/or the presence of aggregates of poorly differentiated mast cells and/or >5 aggregates of >3 mast cells (Krick et al., 2009). For nodal faah inhibitors biopsies, lymph nodes with overt metastases (HN3) were identified by disruption or effacement of normal nodal architecture by discrete foci, nodules, sheets, or masses of neoplastic mast cells (Weishaar et al., 2014). Nodes with early metastasis (HN2) had aggregates of >3 mast cells in sinuses and/or the parenchyma. Pre-metastatic nodes (HN1) had >3 individualized mast cells in sinuses and/or the parenchyma in a minimum of 4 HPF (Weishaar et al., 2014). Non-metastatic nodes had <3 scattered, individualized mast cells in sinuses and/or the parenchyma per HPF. The designation of ‘pre-metastatic’ or ‘early metastasis’ is questionable, since it implies different stages of tumorigenesis rather than an indication of the likelihood of nodal metastasis. In the study by Weishaar et al. (2014), the median survival time for dogs with either HN3 or HN2 nodes was 804 days, whereas dogs with HN1 or HN0 nodes had a median survival time of 1824 days. While these two studies represent a first effort to standardize our interpretation of nodal MCT metastasis, there are also inconsistencies in cytologic versus histologic assessment of nodal spread. One study found a discrepancy between cytologic and histologic evaluation in 20% of examined cases, with a higher percentage of involved lymph nodes being detected histologically (Krick and Kiupel, 2016). While high grade MCTs have a higher likelihood of metastasis to the regional lymph nodes (Krick and Kiupel, 2016), one study reported a mitotic index of ≤5 for 7/12 MCTs with nodal metastases (Worley, 2014). Staging of local lymph nodes should therefore be performed for every cutaneous MCT regardless of grade. PCR to detect activating mutations in c-kit should be considered for both the primary cutaneous MCT and the lymph node metastases for dogs where therapy with TKIs is considered, since it will predict therapeutic response. Future studies need to be directed towards determining the molecular phenotype of mast cells in lymph nodes in order to establish more objective criteria of how to assess lymph node spread.

A promising new tool with great

A promising new tool with great potential to complement classical brucellosis control measures is the identification of animals genetically resistant to Brucella infection. Original studies found pigs and cows naturally resistant to brucellosis (Cameron et al., 1942; Templeton and Adams, 1990); later on, this phenotypic characteristic was linked to polymorphisms at 3′ untranslated region (UTR) of the solute carrier family 11 member A1 (SLC11A1; ex NRAMP1) gene (Barthel et al., 2000; Capparelli et al., 2007). In a previous paper, our group showed that faah inhibitors of the caprine gene SLC11A1 were associated with absence of Brucella-specific antibodies in goats (Iacoboni et al., 2014). Here, we report a new molecular marker in the caprine genome significantly associated with resistance to Brucella infection.
Receptor protein tyrosine phosphatase rho is the most frequently mutated tyrosine phosphatase in human cancer. Several inactivating PTPRT mutations were identified in cancer patients, which include missense, 3′ UTR, and splice site mutations (Zhao et al., 2015), and SNPs variants of PTPRT have been associated with autoimmune diseases, including systemic lupus erythematosus and rheumatoid arthritis (Armstrong et al., 2009). RPTPρ is also a key inhibitor of STAT3 which mediates transcriptional activation in response to several cytokines involved in the inflammatory response, such as IL-6 and IL-2 (Armstrong et al., 2009), but there is no report about the functional importance of RPTPp during bacterial infection.
Alternative splicing variants of PTPRT mRNA were found in mouse and human brain as results of a 57 nucleotides (nt) alternatively spliced region of exon 14, an additional 30 nt alternatively spliced region of exon 16, and an alternatively spliced exon (22a) between exons 22 and 23 (Besco et al., 2001). We may speculate that the haplotype block sequence might affect PTPRT pre-mRNA splicing given its location relative to the 5′ intron/exon border of the eighth intron. Using ESEfinder software (http://krainer01.cshl.edu/cgi-bin/tools/ESE3/esefinder.cgi?process=home) (Cartegni et al., 2003), we observed that the SNP4 was located within overlapping exonic splice enhancer (ESE) motifs, and that the minor allele T abolished potential ESE binding sites for the splicing factors SRSF1 and SRSF6 (Table 5). SRSF1 and SRSF2 are members of the phylogenetically conserved arginine/serine-rich splicing factor (SR) protein family. The SR proteins are essential splicing factors required for constitutive pre-mRNA splicing, and are important regulators of alternative splicing (Long and Caceres, 2009). It was reported that a SNP at intron 8 of bovine CD46 gene causes the elimination of two ESE motif affecting the production of an alternative splicing variant that retains a 48bp intronic sequence (Wang et al., 2014). This data suggest a possible functional role of SNP4 in alternative splicing of caprine PTPRT gene and resistance to Brucella infection. However, the association of SNP4 genotypes with absence or presence of Brucella-specific antibodies should not be considered by definition functional since the polymorphism may only represent an indirect genetic marker of Brucella infection resistance.
Our results are in concordance with a previously reported association between PTPRT polymorphisms and resistance to infection with another facultative intracellular pathogen. Bermingham et al. (2014) identified two PTPRT gene haplotypes blocks associated with resistance to Mycobacterium bovis infection in cattle. Other gene encoded for a protein tyrosine phosphatases (PTP) family member associated with resistance or susceptibility to facultative intracellular bacterial infection is PTPN22, whose product is an intracellular lymphoid-specific phosphatase involved in the inflammatory response with negative regulatory effect on B and T cell activation (Cohen et al., 1999). Gomez et al. (2005) and Boechat et al. (2013) reported an association between PTPN22 gene variants and susceptibility to human tuberculosis; however Bravo et al. (Bravo et al., 2009) failed to associate the C1858T polymorphism in the PTPN22 gene and susceptibility to human brucellosis.

br Equations and practical examples br Note on software

Equations and practical examples

Note on software
All computations were made in the R language and platform (R Core Team 2013) with in-house developed scripts and with the help of the packages Biostrings (Pages et al., 2012), ape (Paradis et al., 2004), seqinr (Charif and Lobry, 2007) and ade4 (Dray and Dufour, 2007). DNA sequences distances were computed with function dna.dist() in package ape, dN and dS were computed with function kaks() in package seqinr using the method of (Li, 1993). An specific R package collecting all developed functions is under preparation.

Acknowledgements
Work in Barcelona supported by grants PI-12/01893, PI13-00456, PI-15/0856 and PI15-00829 from the Spanish Health Ministery. These grants were funded by Instituto de Salud Carlos III and cofinanced by the European Regional Development Fund (ERDF). Work in Madrid supported by grants BFU 2011-23604, SAF 2014-52400-R from Ministerio de Economía y Competitividad, S2013/ABI-2906 (PLATESA-CM) from Comunidad Autónoma de Madrid, and Fundación Ramón Areces. CIBERehd (Centro de Investigación en Red de Enfermedades Hepáticas y Digestivas) is funded by Instituto de Salud Carlos III. C.P. is supported by the Miguel Servet program of the Instituto de Salud Carlos III (CP14/00121).

Introduction
Growth arrest and DNA damage 45 (GADD45) proteins include GADD45A, GADD45B and GADD45G. They are small (around 18kDa) and highly acidic (pH 4.0–4.2) proteins and are mainly located within the nucleus. Human GADD45 proteins share 55–57% homology in their amino faah inhibitors sequences and perform similar functions. Expression of GADD45 can be induced by a variety of stresses including UV-radiation, ionizing irradiation, medium starvation, and methyl methanesulfonate (MMS). This small family of proteins participate in multiple cellular processes such as DNA repair, cell cycle arrest, cell survival and death (Liebermann et al., 2011; Moskalev et al., 2012; Niehrs and Schafer, 2012). These functions of GADD45 proteins involve association with other cellular factors, these include proliferating cell nuclear antigen (PCNA) (Azam et al., 2001; Hall et al., 1995; Smith et al., 1994; Vairapandi et al., 2000), cell division cycle 2 (Cdc2) (Jin et al., 2000; Vairapandi et al., 2002; Zhan et al., 1999), cyclin-dependent kinase inhibitor p21/waf1/cip1 (Fan et al., 1999; Kearsey et al., 1995), stress-responsive kinase MTK1 (Miyake et al., 2007; Takekawa and Saito, 1998), and p38 MARK (Bulavin et al., 2003; Gupta et al., 2006; Salvador et al., 2005). GADD45 proteins also modulate DNA demethylation through association with TBP-associated factor TAF12 and endonuclease XPG (Barreto et al., 2007; Schmitz et al., 2009).
Previous studies have shown that human immunodeficiency virus (HIV)-1 infection induces expression of GADD45 proteins that might contribute to apoptosis (Andersen et al., 2005; Coberley et al., 2004; Genini et al., 2001; Imbeault et al., 2012). In this study, we have investigated whether GADD45 proteins affect HIV-1 replication. Our data showed that overexpression of GADD45 proteins significantly reduced HIV-1 production both in 293T cells and SupT1 cells. We further demonstrated that GADD45 specifically suppressed transcription from HIV-1 long terminal repeat (LTR) promoter. Knockdown of endogenous GADD45 moderately enhanced HIV-1 replication, and activated latent HIV-1. Together, these data suggest GADD45 proteins play a role in controlling HIV-1 transcription and in regulating HIV-1 latency.

Results

Discussion
Here we report that GADD45 proteins exert anti-HIV-1 activity by suppressing viral gene transcription. As opposed to the majority of HIV restriction factors that are interferon-stimulated genes (ISGs), GADD45 proteins are stress inducible. Considering that viral infections also pose stresses to cell, it is not surprising that GADD45 proteins are upregulated during infection by adenovirus (Zhao et al., 2003) and herpes simplex virus 1 (HSV-1) (Khodarev et al., 1999). It has also been reported that HIV-1 infection induces GADD45 expression (Andersen et al., 2005; Coberley et al., 2004; Genini et al., 2001; Imbeault et al., 2012). Although GADD45 expression can be stimulated by different viruses, the mechanisms of stimulation and the effects of increased GADD45 expression differ between viruses. HSV-1 protein ICP0 (Hobbs and DeLuca, 1999) and UL41 (Esclatine et al., 2004a, 2004b) are responsible for GADD45 upregulation, which dysregulates cell cycle and facilitates viral replication. Smith and Mocarski reported that GADD45 proteins interact with the vMIA protein of cytomegalovirus (CMV) and assist vMIA in suppressing cell death (Smith and Mocarski, 2005). On the other hand, hepatitis C virus (HCV) infection diminished GADD45B, which leads to aberrant cell cycle arrest (Higgs et al., 2010). The expression of GADD45A is downregulated by HCV NS5A protein, which subsequently triggers cellular proliferation (Cheng et al., 2013). It is apparent that viruses modulate GADD45 expression and usurp GADD45 to induce or maintain intracellular environments that are favorable for efficient viral replication. However, it remains largely unexplored whether GADD45 proteins influence viral replication. Our data provide the first evidence that expression of GADD45 specifically inhibits HIV-1 production (Fig. 1). Although GADD45 proteins are involved in cell cycle regulation and cell growth, this inhibition is not likely a result of general defect in cell proliferation, because the production of several other retroviruses are not affected by GADD45 overexpression (Fig. 5). We further showed in Fig. S1 that GADD45 overexpression did not trigger cell cycle arrest or apoptosis in 293T cells.

br Fig nbsp xA xA Fig nbsp

Fig. 6 ;  Fig. 7 show the parametric analysis of 42 GHz and 84 GHz operations in terms of power and frequency. These analyses will be very useful in actual operation of the device and the fine tuning can be done to achieve the required power and frequency. The RF power for both the frequencies decreases with increase in magnetic field as appearing in Fig. 6 ;  Fig. 7. This phenomena can be understood by the term detuning factor, which means the faah inhibitors cyclotron motion shifts more and more towards the retarding phase compare to RF phase when magnetic field reduces (electron cyclotron frequency directly depends on magnetic field). In case of beam current, the RF power increases with beam current initially and then become approximately stable. The impact of beam current on the performance of the gyrotron device can be understood easily in term of interaction efficiency (output power/VbIb). The interaction efficiency reduces regularly with the beam current due to the growth in voltage depression [29] ;  [31]. In case of hollow cylindrical beam (as used in gyrotron), voltage depression directly depends on the beam current and thus the interaction efficiency decreases with increase in beam current ( Fig. 6 ;  Fig. 7). Frequency is approximately not affected by magnetic field and beam current.

4. Magnetron injection gun

The beam-wave interaction simulations confirm around 500 kW RF power at 42 GHz and 84 GHz frequencies for the operating modes TE6,2 and TE10,4, respectively. To achieve the required RF power, an electron beam of 1.2 MW power (80 kV, 15 A) must be launched at the first radial maxima of the co-rotating electric field of the selected operating modes, which are 6.2 mm and 7.3 mm for 42 GHz and 84 GHz operations, respectively. A triode type magnetically tunable magnetron injection gun (MT-MIG) is considered and designed for dual frequency operation of gyrotron [18] ;  [25]. In such type of MIG\’s, the electron beam launching position can be controlled by the magnetic field tuning at cathode and interaction cavity. Fig. 8 shows the optimized magnetic field profile for both the frequencies including the cathode position and cavity position.

Fig. 8. Magnetic field profile for 42 GHz and 84 GHz operations.Figure optionsDownload full-size imageDownload high-quality image (203 K)Download as PowerPoint slide

Further, the electrodes geometries are calculated initially by using the trade-off equations derived by Baird and Lawson [42] ;  [43]. In the calculations, beam voltage, beam current, beam launching position and cavity magnetic field are used as the input in an in-house developed code MIGSYN based on trade-off equations [44] ;  [45]. Further, the electrodes geometries are optimized more precisely by using a 2.5 dimensional, FDM trajectory code EGUN [46]. The optimization process is carried out considering the velocity ratio (α), velocity spread and beam launching position as main beam quality parameters. Fig. 9 shows the MIG geometry with the electron beam profile. The gyrating electron beam transports from the emitter to cavity center in quasi laminar flow under the influence of magnetic field profile as shown in Fig. 8. Under the influence of increasing magnetic field, the electrons gyrating motion compressed smoothly (p⊥2/B=constant) and thus the large part of electron beam energy converts into the perpendicular kinetic energy, which leads towards higher beam-wave interaction efficiency. The laminar/quasi laminar flow is necessary for low velocity spread. Finally optimized MIG and electron beam parameters are summarized in Table 4. The electron beam parameters and the beam launching position can be controlled directly by varying the cathode magnetic field and the anode voltage. Fig. 10 ;  Fig. 11 show the sensitivity analyses of velocity ratio and velocity spread with respect to cathode magnetic field and anode voltage for 42 GHz and 84 GHz operations of the device. Such analyses would be helpful in the real fabrication and operation of the device.

Table demonstrates the effect of the score from voting

Table 4 demonstrates the effect of the score (from voting the aligned points on the warping path) in conjunction to that obtained from the DTW algorithm. The sum rule is used for this evaluation. We have experimented over varying code-book sizes (1 to 256 code-vectors) and window widths. To begin, we consider the trend of EER obtained for W=1,W=1, (corresponding to ‘without context’;). Compared to the baseline, with code-books of one to four bits, there is increase in the EER (12.73%, 10.18%, 8.49% and 6.65%), resulting from the coarse quantization process. A small-sized code-book does not adequately capture the finer nuances that discriminate the genuine and forgery signatures, due to their corresponding feature vectors having a tendency to get assigned to the same code-vector. Hence, a higher number of false positives are likely to occur. However, with increasing number of bits in the VQ modeling, the number of code-vectors increase, thus alleviating this problem. Accordingly, for selective choices of M (the code-book size) the EER reduces. We see improvements for M beyond 32. A EER of 5.24% is obtained for a code-book of size 256.
In general, for a given code-book size, we see an improvement in EER with increasing window widths, thus indicating the usefulness of incorporating the contextual information in the DTW formulation. A similar trend of faah inhibitors of EER with increasing code-book size for a given window width can also be inferred. A combination of window width of 7 with code-book size 256 gives the lowest EER of 3.04% (a 52.7% improvement over the baseline system).
To validate on the aforementioned trends, we analyze the mean square distortion values of feature vectors, corresponding to varying number of bits in the VQ model (Table 5). The feature vectors employed are from the five randomly selected enrolled genuine signatures of a user. For small code-book sizes (corresponding to coarse quantization), the mean square error is quite high. However, with finer quantization, achieved with higher-bit/large-sized code-books, the error drops. This in turn could explain the increase in EER, when d2mean scores obtained with small code-book sizes, are fused with the normalized DTW score d1mean.
Though the results of Table 4 pertain to the sum rule combiner, it is worth noting that a similar trend of EERs (with regards to varying code-book size and window widths) is observed for each of the strategies in Table 1. However, owing to space constraints, we present only the best EERs obtained (in Table 6), along with their corresponding values of weight parameter α, code-book size M and window width W. The verification results are quite comparable, with the Weighted Product rule for the parameters (1.4, 256, 11) giving the lowest EER of 2.73%.
We also evaluate the performance of the proposal for varying number of reference signature samples enrolled to the system. The size of the code-book M and window width W used are denoted in the last column of Table 7 for each reference set. The proposed system verifies the authenticity of a user quite effectively with as low as two reference signatures.