Object detection using feature subset selection

Past work on object detection has emphasized the issues of feature extraction and classification, however, relatively less attention has been given to the critical issue of feature selection. The main trend in feature extraction has been representing the data in a lower dimensional space, for example, using principal component analysis (PCA). Without using an effective scheme to select an appropriate set of features in this space, however, these methods rely mostly on powerful classification algorithms to deal with redundant and irrelevant features. In this paper, we argue that feature selection is an important problem in object detection and demonstrate that genetic algorithms (GAs) provide a simple, general, and powerful framework for selecting good subsets of features, leading to improved detection rates. As a case study, we have considered PCA for feature extraction and support vector machines (SVMs) for classification. The goal is searching the PCA space using GAs to select a subset of eigenvectors encoding important information about the target concept of interest. This is in contrast to traditional methods selecting some percentage of the top eigenvectors to represent the target concept, independently of the classification task. We have tested the proposed framework on two challenging applications: vehicle detection and face detection. Our experimental results illustrate significant performance improvements in both cases.     

Graduate training of scientist–practitioner-oriented clinical psychologists: Where we can improve.

To improve psychology graduate programs, the training provided by scientist- or practitioner-oriented graduate programs in clinical psychology is evaluated from the perspective of a director of clinical training in an American Psychological Association-approved internship. Areas in which graduate training may be improved are divided into knowledge, supervision, research, and professional development. Specific training deficits are enumerated, and suggestions for their remediation are offered. (15 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Multiscale analysis of stiffness and fracture of nanoparticle-reinforced composites using micromechanics and global–local finite element models

A combined micromechanics analysis and global–local finite element method is proposed to study the interaction of particles and matrix at the nano-scale near a crack tip. An analytical model is used to obtain the effective elastic modulus of nanoparticle-reinforced composites, then a global–local multi-scale finite element model with effective homogeneous material properties is used to study the fracture of a compact tension sample. For SiO₂ particle-reinforced epoxy composites with various volume fractions, the simulation results for effective elastic modulus, fracture toughness, and critical strain energy release rate show good agreement with previously published experimental data. It is demonstrated that the proposed parametric multi-scale model can be used to efficiently study the toughness mechanisms at both the macro and nano-scale.     

The P2X4 Receptor: Cellular and Molecular Characteristics of a Promising Neuroinflammatory Target

The adenosine 5′-triphosphate-gated P2X4 receptor channel is a promising target in neuroinflammatory disorders, but the ability to effectively target these receptors in models of neuroinflammation has presented a constant challenge. As such, the exact role of P2X4 receptors and their cell signalling mechanisms in human physiology and pathophysiology still requires further elucidation. To this end, research into the molecular mechanisms of P2X4 receptor activation, modulation, and inhibition has continued to gain momentum in an attempt to further describe the role of P2X4 receptors in neuroinflammation and other disease settings. Here we provide an overview of the current understanding of the P2X4 receptor, including its expression and function in cells involved in neuroinflammatory signalling. We discuss the pharmacology of P2X4 receptors and provide an overview of P2X4-targeting molecules, including agonists, positive allosteric modulators, and antagonists. Finally, we discuss the use of P2X4 receptor modulators and antagonists in models of neuroinflammatory cell signalling and disease.     

Heavy metal precipitation by polycation-polyanion complex of PEI and its phosphonomethylated derivative

The removal of various heavy metal ions such as Cu2+, Co2+, Zn2+, Ni2+ and Pb2+ from aqueous solutions by precipitation with polyelectrolyte complex of PPEI and PEI was conducted. Heavy metal binding with PPEI was initially allowed to occur and then upon equilibration, PEI was added to initiate precipitation of the polyelectrolyte complex together with the heavy metal ion. The PPEI-PEI system was found effective for heavy metal scavenging purposes even in the presence of high concentrations of non-transition metal ions like Na+. Heavy metal concentration may be reduced beyond emission standards for industrial wastewaters. The PPEI-PEI polyelectrolyte complex was found to be more effective than traditional precipitation methods for the treatment of a representative electroless Ni plating waste solution.     

Design of efficient, broadband single-element (20-80 MHz) ultrasonic transducers for medical imaging applications

This paper discusses the design, fabrication, and testing of sensitive broadband lithium niobate (LiNbO/sub 3/) single-element ultrasonic transducers in the 20-80 MHz frequency range. Transducers of varying dimensions were built for an f# range of 2.0-3.1. The desired focal depths were achieved by either casting an acoustic lens on the transducer face or press-focusing the piezoelectric into a spherical curvature. For designs that required electrical impedance matching, a low impedance transmission line coaxial cable was used. All transducers were tested in a pulse-echo arrangement, whereby the center frequency, bandwidth, insertion loss, and focal depth were measured. Several transducers were fabricated with center frequencies in the 20-80 MHz range with the measured -6 dB bandwidths and two-way insertion loss values ranging from 57 to 74% and 9.6 to 21.3 dB, respectively. Both transducer focusing techniques proved successful in producing highly sensitive, high-frequency, single-element, ultrasonic-imaging transducers. In vivo and in vitro ultrasonic backscatter microscope (UBM) images of human eyes were obtained with the 50 MHz transducers. The high sensitivity of these devices could possibly allow for an increase in depth of penetration, higher image signal-to-noise ratio (SNR), and improved image contrast at high frequencies when compared to previously reported results.     

Development of an Active Thermoplastic Film with Oxygen Scavengers Made of Activated Carbon and Sodium Erythorbate

An active thermoplastic film made of low‐density polyethylene (LDPE) filled with oxygen scavengers made of powdered activated carbon (PAC) impregnated with sodium erythorbate (SE) was developed for packaging applications. Initial tests indicated that the impregnation of PAC with SE enhanced the heat resistance of SE, thereby allowing processing at temperatures typical of LDPE manufacturing. Subsequently, LDPE films with PAC/SE particles were manufactured in coupons that represented a typical juice package, and experiments indicated that these films absorbed 3.57 mg of oxygen in 11 days. This amount corresponded to 80% the concentration of oxygen in the headspace of the package. Furthermore, findings indicated that active particles alone have 10 times higher oxygen absorption capacity than the active LDPE film. Finally, the physical properties of the film were characterized by microscopy where oxygen scavengers showed a good dispersion within the matrix. However, 20 wt.% of these active particles decreased tensile strength of the film by 53%. Copyright © 2014 John Wiley & Sons, Ltd.     

Progress in CdZnTe substrate producibility and critical drivers of IRFPA yield originating with CdZnTe substrates

With the goal of maximizing the yield of infrared focal plane arrays (IRFPAs), Santa Barbara Research Center’s (SBRC) Infrared Materials Producibility Program (IRMP) has focused on assessing and improving the quality, yield, and throughput of CdZnTe substrates. A baseline detector lot was fabricated to identify the critical drivers of IRFPA yield coming from the substrates and to evaluate the quality and yield of the current vendor base for CdZnTe substrates. Substrate induced defects and impurities that can potentially affect device performance and operability were carefully mapped out in detail on 44 × 67 mm² size substrates, received from IRMP substrate vendors as well as SBRC. This paper will report on the correlations found between this substrate characterization data base and the IRFPA level defect distributions. Key results from these correlation studies are: (1) extended defects found on the substrates with the Nakagawa etch correlated well with responsivity reduction in the final IRFPA; (2) cross-hatch patterns that were evident in the responsivity map correlated well with similar features seen by x-ray topography on LPE double layers; and (3) a possible correlation of device performance (leakage current at 78K) with copper and lithium impurities in the substrate. Recent initiatives toward improving the quality and yield of the substrate growth process have focused on improving purity in the pre-growth charge preparation, modification of growth parameters to reduce defects and scaling up of the vertical Bridgman growth process from its current 67 mm diameter boule size to 92 mm diameter boules. Promising initial results from the large diameter boule growth process will be shown. The 92 mm diameter CdZnTe boule (6 kg charge) shows two predominant single crystal grains encompassing 75% of boule volume. Defect characterization of boules grown under baseline and modified conditions is discussed.     

Endometrial Epithelial ARID1A Is Required for Uterine Immune Homeostasis during Early Pregnancy

A growing body of work suggests epigenetic dysregulation contributes to endometriosis pathophysiology and female infertility. The chromatin remodeling complex subunit AT-rich interaction domain 1A (ARID1A) must be properly expressed to maintain normal uterine function. Endometrial epithelial ARID1A is indispensable for pregnancy establishment in mice through regulation of endometrial gland function; however, ARID1A expression is decreased in infertile women with endometriosis. We hypothesized that ARID1A performs critical operations in the endometrial epithelium necessary for fertility besides maintaining gland function. To identify alterations in uterine gene expression resulting from loss of epithelial ARID1A, we performed RNA-sequencing analysis on pre-implantation uteri from Ltf^iCre/+Arid1a^f/f and control mice. Differential expression analysis identified 4181 differentially expressed genes enriched for immune-related ingenuity canonical pathways including agranulocyte adhesion and diapedesis and natural killer cell signaling. RT-qPCR confirmed an increase in pro-inflammatory cytokine and macrophage-related gene expression but a decrease in natural killer cell signaling. Immunostaining confirmed a uterus-specific increase in macrophage infiltration. Flow cytometry delineated an increase in inflammatory macrophages and a decrease in uterine dendritic cells in Ltf^iCre/+Arid1a^f/f uteri. These findings demonstrate a role for endometrial epithelial ARID1A in suppressing inflammation and maintaining uterine immune homeostasis, which are required for successful pregnancy and gynecological health.