Study of the Mechanical Properties of Wood under Transverse Compression Using Monto Carlo Simulation-Based Stochastic FE Analysis

Wood is an anisotropic material, the mechanical properties of which are strongly influenced by its microstructure. In wood, grain compression strength and modulus are the weakest perpendicular to the grain compared to other grain directions. FE (finite element) models have been developed to investigate the mechanical properties of wood under transverse compression. However, almost all existing models were deterministic. Thus, the variations of geometry of the cellular structure were not considered, and the statistical characteristic of the mechanical property was not involved. This study aimed to develop an approach to investigate the compression property of wood in a statistical sense by considering the irregular geometry of wood cells. First, the mechanical properties of wood under radial perpendicular to grain compression was experimentally investigated, then the statistical characteristic of cell geometry was extracted from test data. Finally, the mechanical property of wood was investigated using the finite element method in combination with the Monte Carlo Simulation (MCS) techniques using randomly generated FE models. By parameter sensitivity analysis, it was found that the occurrence of the yield points was caused by the bending or buckling of the earlywood axial tracheid cell wall in the tangential direction. The MCS-based stochastic FE analysis was revealed as an interesting approach for assessing the micro-mechanical performance of wood and in assisting in understanding the mechanical behavior of wood based on its hierarchical structure.

Clinical and statistical characteristic of references at the third stage of ophthalmic care

AIM:to carry out a clinical and statistical analysis of patients references at the third stage of ophthalmic care in Azerbaijan Republic conditions. MATERIALS AND METHODS:Materials of the National Ophthalmology Center named after the Academician Zarifa Aliyeva are used. All the cases of primary references during 2019 are analyzed. RESULTS:The lowest proportion of unjustified references was among residents of republican subordination cities (1.3 0.1%), where the frequency of primary visits is low as well (1.28 0.04). The rate of unjustified references of regional centre city residents (4.8 0.1%), and of those of rural settlements (10.7 0.1%) were significantly different. There were significant differences concerning the rate of primary visits (2.32 0.026 and 2.56 0.024). The rate of primary visits between male and female populations also was significantly different (2.58 0.024 and 3.02 0.021,р 0.01). CONCLUSION:The rate of primary references at the third stage was 4.62 0.04 in Baku, 2.56 0.024 in rural settlements, 2.32 0.031 in regional centre cities, and 1.28 0.04 in republican subordination cities; it had significant gender specificities. Within primary visit causes, accommodation and refraction disorders (32.3 0.5%), ocular trauma and that of eye adnexa (19.7 0.4%) prevail.

The statistical analysis of random-valued impulse noise detection techniques based on the local image characteristic: ROAD, ROLD and RORD

<p>Advances in local image statistical analysis have made possible the random-valued impulse noise detection but the current noise detections based on ROAD (Rank-Ordered Absolute Differences), ROLD (Rank-Ordered Logarithmic Differences) and RORD (Rank-Ordered Relative Differences), which are the most three effective and practical detections using the local image statistical characteristic, operates effectively on different noise density and different image statistical characteristic. To address these issues, this paper proposes the comparative analysis on the noise detections based on ROAD, ROLD and RORD. Therefore, the first contribution is the comparative statistical distribution of these three noise detections. By comprehensive experiment at each noise density, the optimized detected threshold is later determined from four benchmark data: Lena, Girl, Pepper and Airplane. Moreover, the maximum detection accuracy for each case is comparatively demonstrated by using the noise detections based on ROAD, ROLD and RORD with the optimized detected threshold.</p>

Utilizing Class Information for Deep Network Representation Shaping

Statistical characteristics of deep network representations, such as sparsity and correlation, are known to be relevant to the performance and interpretability of deep learning. When a statistical characteristic is desired, often an adequate regularizer can be designed and applied during the training phase. Typically, such a regularizer aims to manipulate a statistical characteristic over all classes together. For classification tasks, however, it might be advantageous to enforce the desired characteristic per class such that different classes can be better distinguished. Motivated by the idea, we design two class-wise regularizers that explicitly utilize class information: class-wise Covariance Regularizer (cw-CR) and classwise Variance Regularizer (cw-VR). cw-CR targets to reduce the covariance of representations calculated from the same class samples for encouraging feature independence. cw-VR is similar, but variance instead of covariance is targeted to improve feature compactness. For the sake of completeness, their counterparts without using class information, Covariance Regularizer (CR) and Variance Regularizer (VR), are considered together. The four regularizers are conceptually simple and computationally very efficient, and the visualization shows that the regularizers indeed perform distinct representation shaping. In terms of classification performance, significant improvements over the baseline and L1/L2 weight regularization methods were found for 21 out of 22 tasks over popular benchmark datasets. In particular, cw-VR achieved the best performance for 13 tasks including ResNet-32/110.