Search

The effect of astigmatism on laser air filamentation was studied numerically and experimentally. Limited supply of energy to the filament is shown to be the main feature of aberrational focusing, which facilitated the organization of quasi-solitons in the post-filament region. Quasi-solitons are realized due to the balance between linear diffraction and Kerr nonlinearity in the presence of background environment around zones with increased intensity. A highly directional SC of visible light is formed outside the visible filament zone in the resulting spatial quasi-solitons.

GaSe crystals with 0, 0.07, 0.38, 0.67, 2.07, 3, and 5 wt % Te are grown. GaSe:Te (≤2.07 wt %) crystals are suitable for nonlinear optics applications. The optimum doping level is 0.07 wt %. This minimizes optical losses and increases the damage threshold by ≥20%, hardness by 33%, and the efficiency of femtosecond Ti:Sa laser frequency conversion into the 8.8–24 μm and 0.2–3 THz ranges by ≥50%.

The results of formation conditions studies of a highly directional supercontinuum (SC) in a visible spectrum range obtained upon aberration spherical-mirror focusing of a radiation pulse with a wavelength of 940 nm, duration of 70 fs, and energy of 8–15 mJ are presented. It is shown that after visible filament there are two directed white light beams diverging relative to each other at an angle of 1.4 0 . Formation every light beam occurs through a gradual conversion of the spectral composition from long wavelength to short wavelength (to 350 nm) in a spatially stable structure similar to a soliton with a transverse dimension ≤ 300 μm. The nature of the appearance these beams is due to formation of two zones with higher intensity before meridional plate owing to the distortion of the wavefront of the laser beam in conditions of the astigmatism and the Kerr effect. In result two minima in the phase distribution located outside the beam axis are realized, which lead to the appearance of two off-axis areas with higher radiation intensity and as a consequence of this the formation of two highly directional laser beams.

Centimeter-sized crystals of GaSe have been grown by a modified vertical Bridgman method from a melt with content 0.01, 0.05, 0.1, 0.2, 0.5, 1, and 2 at.% Al and 0.025, 0.1, 0.5, 1, and 2 at.% Er using heat field rotation. Their physical properties have been investigated and compared with the properties of doped crystals with content 0.1, 0.5, 1, 2, 3, 5, 7, and 10.2 mass% S, 0.01, 0.1, 0.5, 1, 2, 3, and 5 mass% In, and 0.01, 0.1, 0.5, 1, and 2 mass% Te, grown by the same method and by the conventional Bridgman method. Optimal levels of doping of GaSe crystals with Al and Er have been established here for the first time to be equal to 0.01–0.05 at.% Al and ~0.5 at.% Er, respectively, of the growth charge.

With most image steganalysis traditionally based on supervised machine learning methods, the size of training data has remained static at up to 20000 training examples. This potentially leads to the classifier being undertrained for larger feature sets and it may be too narrowly focused on characteristics of a source of cover images, resulting in degradation in performance when the testing source is mismatched or heterogeneous. However it is not difficult to obtain larger training sets for steganalysis through simply taking more photos or downloading additional images. Here, we investigate possibilities for creating steganalysis classifiers trained on large data sets using large feature vectors. With up to 1.6 million examples, naturally simpler classification engines must be used and we examine the hypothesis that simpler classifiers avoid overtraining and so perform better on heterogeneous data. We highlight the possibilities of online learners, showing that, when given sufficient training data, they can match or exceed the performance of complex classifiers such as Support Vector Machines. This applies to both their accuracy and training time. We include some experiments, not previously reported in the literature, which provide benchmarks of some known feature sets and classifier combinations.

The security of a steganography system is defined by our ability to detect it. It is of no surprise then that steganography and steganalysis both depend heavily on the accuracy and robustness of our detectors. This is especially true when real-world data is considered, due to its heterogeneity. The difficulty of such data manifests itself in a penalty that has periodically been reported to affect the performance of detectors built on binary classifiers; this is known as cover source mismatch. It remains unclear how the performance drop that is associated with cover source mismatch is mitigated or even measured. In this thesis we aim to show a robust methodology to empirically measure its effects on the detection accuracy of steganalysis classifiers. Some basic machine-learning based methods, which take their origin in domain adaptation, are proposed to counter it. Specifically, we test two hypotheses through an empirical investigation. First, that linear classifiers are more robust than non-linear classifiers to cover source mismatch in real-world data and, second, that linear classifiers are so robust that given sufficiently large mismatched training data they can equal the performance of any classifier trained on small matched data. With the help of theory we draw several nontrivial conclusions based on our results. The penalty from cover source mismatch may, in fact, be a combination of two types of error; estimation error and adaptation error. We show that relatedness between training and test data, as well as the choice of classifier, both have an impact on adaptation error, which, as we argue, ultimately defines a detector's robustness. This provides a novel framework for reasoning about what is required to improve the robustness of steganalysis detectors. Whilst our empirical results may be viewed as the first step towards this goal, we show that our approach provides clear advantages over earlier methods. To our knowledge this is the first study of this scale and structure.

Modern steganalysis can be incredibly sensitive and accurate, but only in an artificial setting in which the training data is from the exact same image source as the data being tested. When faced with mismatched training data, performance degrades, often substantially. Some recent publications have advocated the use of very simple classifiers for steganalysis. It is folklore from the machine learning literature that simple classifiers may be more robust to variations between training and testing data, and this paper investigates whether this is true for steganalysis also. Simple classifiers are compared with the classic complex Kernel Support Vector Machine, faced with training data which is mismatched with the testing data. A real-world source of images is used.

The aim of this study was to assess changes in child behavior and mother-child emotional availability after mother's participation in a parent training program, and to assess the role of various factors which might explain individual variability in these changes. After mothers had participated in the parent training program they reported less child externalizing behavior problems and less child internalizing behavior problems. The child's preschool teacher also reported decreased child externalizing problems during the post-training assessment measure. Observational measures showed increased mother-child emotional availability. Post-training behavioral differences were found depending upon the child's age, with greater changes in positive mother-child emotional availability and preschool teachers reports of decreased internalizing behavior for the older group of children, 4 -5 years old. A decrease in child internalizing behavior was more prominent in families with more than one child.

Objectives The aim of this study was to compare the pharmacodynamics of telavancin (TLV) and vancomycin (VAN) with Staphylococcus aureus. Their concentrations were simulated between the MIC and the mutant prevention concentration (MPC), and above the MPC. Methods Two strains of S. aureus, glycopeptide-intermediate S. aureus (GISA) Mu-50 and ATCC 43300, were exposed for 5 days to once-daily TLV (half-life 8 h) and twice-daily VAN (half-life 6 h). The simulated ratios of 24 h area under the curve (AUC(24)) to MIC varied from 30-50 to 3400 h. The cumulative antimicrobial effect was expressed by ABBC (area between the level corresponding to the starting inoculum and the time-kill curve calculated from time 0 to 144 h). Results With each antibiotic, the ABBC versus log AUC(24)/MIC relationships were bacterial strain-independent. A sigmoid model fits combined data on both organisms exposed to TLV (r(2)=0.78) or VAN (r(2)=0.85). Comparable effects of the proposed therapeutic dose of TLV (10 mg/kg) and a clinical dose of VAN (2x1 g) were predicted for MRSA ATCC 43300 (AUC(24)/MIC 3400 and 500 h, respectively) and a 1.6-fold greater effect of TLV for GISA Mu-50 compared with VAN (AUC(24)/MIC 1700 and 130 h, respectively). Mutants of S. aureus ATCC 43300 resistant to 2x and 4x MIC of VAN but not TLV were enriched in these simulations. No selection of TLV- and VAN-resistant mutants of GISA Mu-50 was observed. Conclusions These in vitro data suggest that the effects of clinically attainable AUC/MIC ratios of TLV are similar to those of VAN on S. aureus 43300 and 2-fold greater on GISA Mu-50.

Objectives: To relate the enrichment of linezolid-resistant Enterococcus faecium with linezolid pharmacokinetics, the pharmacodynamics of linezolid and its ability to prevent the selection of resistant mutants were studied in an in vitro model that simulates antibiotic concentrations in and out of the mutant selection window (MSW), i.e. the concentration range from the MIC to the mutant prevention concentration (MPC). Methods: A clinical isolate of E. faecium (MIC 1.8mg/L and MPC 7mg/L) at a starting inoculum of 8 log cfu/mL was exposed to twice-daily linezolid, alone and in combination with once-daily doxycycline (MIC 0.2 mg/L and MPC 3.4 mg/L), for 3 consecutive days in a hollow-fibre two-compartment model. Results: The ratios of 24 h area under the curve (AUC 24 ) to MIC of linezolid were estimated at 70, 100 and 230 h and those of doxycycline were estimated at 230 and 720 h. At the two lower AUC 24/ MIC ratios of linezolid given alone, E. faecium resistant to 2 x MIC-16 x MIC and 2 x MIC-8 x MIC of linezolid, respectively, were selectively enriched with a concomitant slight loss in susceptibility. Neither growth on linezolid-containing media nor changes in susceptibility occurred at the high AUC 24 /MIC ratio. A similar protective effect was observed with the minimal AUC 24 /MIC ratio of linezolid (70 h) combined with doxycycline at an AUC 24 /MIC of 230 h. Conclusions: This study suggests that selection of linezolid-resistant enterococci can be predicted from the MSW concept and can be prevented by linezolid given in combination with doxycycline, each at suboptimal AUC 24 /MIC ratios.