State estimation schemes for independent component coupled hidden Markov models

Conventional Hidden Markov models generally consist of a Markov chain observed through a linear map corrupted by additive noise. This general class of model has enjoyed a huge and diverse range of applications, for example, speech processing, biomedical signal processing and more recently quantitative finance. However, a lesser known extension of this general class of model is the so-called Factorial Hidden Markov Model (FHMM). FHMMs also have diverse applications, notably in machine learning, artificial intelligence and speech recognition [13, 17]. FHMMs extend the usual class of HMMs, by supposing the partially observed state process is a finite collection of distinct Markov chains, either statistically independent or dependent. There is also considerable current activity in applying collections of partially observed Markov chains to complex action recognition problems, see, for example, [6]. In this article we consider the Maximum Likelihood (ML) parameter estimation problem for FHMMs. Much of the extant literature concerning this problem presents parameter estimation schemes based on full data log-likelihood EM algorithms. This approach can be slow to converge and often imposes heavy demands on computer memory. The latter point is particularly relevant for the class of FHMMs where state space dimensions are relatively large. The contribution in this article is to develop new recursive formulae for a filter-based EM algorithm that can be implemented online. Our new formulae are equivalent ML estimators, however, these formulae are purely recursive and so, significantly reduce numerical complexity and memory requirements. A computer simulation is included to demonstrate the performance of our results. © Taylor & Francis Group, LLC.

The effect of shear strength on the ballistic response of laminated composite plates

The ballistic performance of clamped circular carbon fibre reinforced polymer (CFRP) and Ultra High Molecular Weight Polyethylene (UHMWPE) fibre composite plates of equal areal mass and 0/90 lay-up were measured and compared with that of monolithic 304 stainless steel plates. The effect of matrix shear strength upon the dynamic response was explored by testing: (i) CFRP plates with both a cured and uncured matrix and (ii) UHMWPE laminates with identical fibres but with two matrices of different shear strength. The response of these plates when subjected to mid-span, normal impact by a steel ball was measured via a dynamic high speed shadow moiré technique. Travelling hinges emanate from the impact location and travel towards the supports. The anisotropic nature of the composite plate results in the hinges travelling fastest along the fibre directions and this results in square-shaped moiré fringes in the 0/90 plates. Projectile penetration of the UHMWPE and the uncured CFRP plates occurs in a progressive manner, such that the number of failed plies increases with increasing velocity. The cured CFRP plate, of high matrix shear strength, fails by cone-crack formation at low velocities, and at higher velocities by a combination of cone-crack formation and communition of plies beneath the projectile. On an equal areal mass basis, the low shear strength UHMWPE plate has the highest ballistic limit followed by the high matrix shear strength UHMWPE plate, the uncured CFRP, the steel plate and finally the cured CFRP plate. We demonstrate that the high shear strength UHMWPE plate exhibits Cunniff-type ballistic limit scaling. However, the observed Cunniff velocity is significantly lower than that estimated from the laminate properties. The data presented here reveals that the Cunniff velocity is limited in its ability to characterise the ballistic performance of fibre composite plates as this velocity is independent of the shear properties of the composites: the ballistic limit of fibre composite plates increases with decreasing matrix shear strength for both CFRP and UHMWPE plates. © 2013 Elsevier Masson SAS. All rights reserved.

Optimization on the orthogonal group for independent component analysis

This paper derives a new algorithm that performs independent component analysis (ICA) by optimizing the contrast function of the RADICAL algorithm. The core idea of the proposed optimization method is to combine the global search of a good initial condition with a gradient-descent algorithm. This new ICA algorithm performs faster than the RADICAL algorithm (based on Jacobi rotations) while still preserving, and even enhancing, the strong robustness properties that result from its contrast. © Springer-Verlag Berlin Heidelberg 2007.

Geometric optimization methods for independent component analysis applied on gene expression data

DNA microarrays provide a huge amount of data and require therefore dimensionality reduction methods to extract meaningful biological information. Independent Component Analysis (ICA) was proposed by several authors as an interesting means. Unfortunately, experimental data are usually of poor quality- because of noise, outliers and lack of samples. Robustness to these hurdles will thus be a key feature for an ICA algorithm. This paper identifies a robust contrast function and proposes a new ICA algorithm. © 2007 IEEE.

Gate commutated thyristor with voltage independent maximum controllable current

In this letter, we use a novel 3-D model, earlier calibrated with experimental results on standard gate commutated thyristors (GCTs), with the aim to explain the physics behind the high-power technology (HPT) GCT, to investigate what impact this design would have on 24 mm diameter GCTs, and to clarify the mechanisms that limit safe switching at different dc-link voltages. The 3-D simulation results show that the HPT design can increase the maximum controllable current in 24 mm diameter devices beyond the realm of GCT switching, known as the hard-drive limit. It is proposed that the maximum controllable current becomes independent of the dc-link voltage for the complete range of operating voltage. © 1980-2012 IEEE.

Cognitive Tomography Reveals Complex, Task-Independent Mental Representations

Humans develop rich mental representations that guide their behavior in a variety of everyday tasks. However, it is unknown whether these representations, often formalized as priors in Bayesian inference, are specific for each task or subserve multiple tasks. Current approaches cannot distinguish between these two possibilities because they cannot extract comparable representations across different tasks [1-10]. Here, we develop a novel method, termed cognitive tomography, that can extract complex, multidimensional priors across tasks. We apply this method to human judgments in two qualitatively different tasks, "familiarity" and "odd one out," involving an ecologically relevant set of stimuli, human faces. We show that priors over faces are structurally complex and vary dramatically across subjects, but are invariant across the tasks within each subject. The priors we extract from each task allow us to predict with high precision the behavior of subjects for novel stimuli both in the same task as well as in the other task. Our results provide the first evidence for a single high-dimensional structured representation of a naturalistic stimulus set that guides behavior in multiple tasks. Moreover, the representations estimated by cognitive tomography can provide independent, behavior-based regressors for elucidating the neural correlates of complex naturalistic priors. © 2013 The Authors.

Independent Transport and Sorting of Functionally Distinct Protein Families in Tetrahymena thermophila Dense Core Secretory Granules

Dense core granules (DCGs) in Tetrahymena thermophila contain two protein classes. Proteins in the first class, called granule lattice (Grl), coassemble to form a crystalline lattice within the granule lumen. Lattice expansion acts as a propulsive mechanism during DCG release, and Grl proteins are essential for efficient exocytosis. The second protein class, defined by a C-terminal beta/gamma-crystallin domain, is poorly understood. Here, we have analyzed the function and sorting of Grt1p (granule tip), which was previously identified as an abundant protein in this family. Cells lacking all copies of GRT1, together with the closely related GRT2, accumulate wild-type levels of docked DCGs. Unlike cells disrupted in any of the major GRL genes, Delta GRT1 Delta GRT2 cells show no defect in secretion, indicating that neither exocytic fusion nor core expansion depends on GRT1. These results suggest that Grl protein sorting to DCGs is independent of Grt proteins. Consistent with this, the granule core lattice in Delta GRT1 Delta GRT2 cells appears identical to that in wild-type cells by electron microscopy, and the only biochemical component visibly absent is Grt1p itself. Moreover, gel filtration showed that Grl and Grt proteins in cell homogenates exist in nonoverlapping complexes, and affinity-isolated Grt1p complexes do not contain Grl proteins. These data demonstrate that two major classes of proteins in Tetrahymena DCGs are likely to be independently transported during DCG biosynthesis and play distinct roles in granule function. The role of Grt1p may primarily be postexocytic; consistent with this idea, DCG contents from Delta GRT1 Delta GRT2 cells appear less adhesive than those from the wild type.

The synthesis of astaxanthin esters, independent of the formation of cysts, highly correlates with the synthesis of fatty acids in Haematococcus pluvialis

The compositions and contents of astaxanthin esters and fatty acids in four types of Haematococcus pluvialis cells were studied by HPLC and GC-MS. Results showed that the synthesis and accumulation of astaxanthin was independent of the formation of cysts, but was highly correlated with the synthesis and accumulation of fatty acids, though it is an well known phenomenon that the accumulation of astaxanthin is usually accompanied by the formation of cyst. The red cysts contain more than 30% of fatty acids, with 81% of the unsaturated fatty acids. Taken together, besides a resource of astaxanthin, H. pluvialis would be a good resource of valuable fatty acids.

Independent adaptation to riverine habitats allowed survival of ancient cetacean lineages

The four species of "river dolphins" are associated with six separate great river systems on three subcontinents and have been grouped for more than a century into a single taxon based on their similar appearance. However, several morphologists recently questioned the monophyly of that group. By using phylogenetic analyses of nucleotide sequences from three mitochondrial and two nuclear genes, we demonstrate with statistical significance that extant river dolphins are not monophyletic and suggest that they are relict species whose adaptation to riverine habitats incidentally insured their survival against major environmental changes in the marine ecosystem or the emergence of Delphinidae.

Design and fabrication of polarization-independent micro-ring resonators

Origin of polarization sensitivity of photonic wire waveguides (PWWs) is analysed and the effective refractive indices of two different polarization states are calculated by the three-dimensional full-vector beam propagation method. We find that PWWs are polarization insensitive if the distribution of its refractive index is uniform and the cross section is square. An MRR based on such a polarization-insensitive PWW is fabricated on an 8-inch silicon-on-insulator wafer using 248-nm deep ultraviolet lithography and reactive ion etching. The quasi-TE mode is resonant at 1542.25 nm and 1558.90 nm, and the quasi-TM mode is resonant at 1542.12 nm and 1558.94 nm. The corresponding polarization shift is 0.13 nm at the shorter wavelength and 0.04 nm at the longer wavelength. Thus the fabricated device is polarization independent. The extinction ratio is larger than 10 dB. The 3 dB bandwidth is about 2.5 nm and the Qvalue is about 620 at 1558.90 nm.

Temperature-independent fiber Bragg grating strain sensor using bimetal cantilever

A new packaged fiber Bragg grating using bimetal cantilever beam as the strain agent is presented. The grating is two-point attached on one specific surface of the bimetal beam which consists of two metallic material with different thermal-expansion coefficient. Thereby the grating can be compressed or stretched along with the cantilever beam while temperature varies and temperature compensation can be realized. At the same time, grating chirping can be avoided for the particular attaching method. Experiment results demonstrated that the device is able to automatically compensate temperature induced wavelength shift. The temperature dependence of Bragg wavelength reduced to -0.4 pm/degrees C over the temperature range from -20 to 60 degrees C. This fiber grating package technique is cost effective and can be used in strain sensing. (c) 2005 Elsevier Inc. All rights reserved.

Temperature-independent FBG pressure sensor with high sensitivity

This study develops a fiber Bragg grating (FBG) pressure sensor partly shielded with a metal tube. The thermal-strain cross effect is avoided and its pressure sensitivity is increased to -2.44 x 10(-3) MPa, about 1200 times as that of a bare fiber grating. Due to its good sensing linearity, this sensor can be applied in the measurement of hydraulic pressure and vibration. (c) 2006 Elsevier Inc. All rights reserved.

Information geometry theory of high-dimension space and application for speaker independent continuous digit speech recognition

In the light of descriptive geometry and notions in set theory, this paper re-defines the basic elements in space such as curve and surface and so on, presents some fundamental notions with respect to the point cover based on the High-dimension space (HDS) point covering theory, finally takes points from mapping part of speech signals to HDS, so as to analyze distribution information of these speech points in HDS, and various geometric covering objects for speech points and their relationship. Besides, this paper also proposes a new algorithm for speaker independent continuous digit speech recognition based on the HDS point dynamic searching theory without end-points detection and segmentation. First from the different digit syllables in real continuous digit speech, we establish the covering area in feature space for continuous speech. During recognition, we make use of the point covering dynamic searching theory in HDS to do recognition, and then get the satisfying recognized results. At last, compared to HMM (Hidden Markov models)-based method, from the development trend of the comparing results, as sample amount increasing, the difference of recognition rate between two methods will decrease slowly, while sample amount approaching to be very large, two recognition rates all close to 100% little by little. As seen from the results, the recognition rate of HDS point covering method is higher than that of in HMM (Hidden Markov models) based method, because, the point covering describes the morphological distribution for speech in HDS, whereas HMM-based method is only a probability distribution, whose accuracy is certainly inferior to point covering.

A speaker-independent continuous speech recognition system using biomimetic pattern recognition

In speaker-independent speech recognition, the disadvantage of the most diffused technology (HMMs, or Hidden Markov models) is not only the need of many more training samples, but also long train time requirement. This paper describes the use of Biomimetic pattern recognition (BPR) in recognizing some mandarin continuous speech in a speaker-independent manner. A speech database was developed for the course of study. The vocabulary of the database consists of 15 Chinese dish's names, the length of each name is 4 Chinese words. Neural networks (NNs) based on Multi-weight neuron (MWN) model are used to train and recognize the speech sounds. The number of MWN was investigated to achieve the optimal performance of the NNs-based BPR. This system, which is based on BPR and can carry out real time recognition reaches a recognition rate of 98.14% for the first option and 99.81% for the first two options to the persons from different provinces of China speaking common Chinese speech. Experiments were also carried on to evaluate Continuous density hidden Markov models (CDHMM), Dynamic time warping (DTW) and BPR for speech recognition. The Experiment results show that BPR outperforms CDHMM and DTW especially in the cases of samples of a finite size.