Foramen Vesalius is a small, variable and an inconstant foramen located in the greater wing of sphenoid, anteromedial to the foramen ovale, in the middle cranial fossa. This foramen is also known as emissary sphenoidal foramen as it transmits a small emissary vein which drains Cavernous sinus. The importance of this foramen lies in the fact that an infected thrombus from an extracranial source may reach cavernous sinus. The present study was undertaken to observe the incidence of foramen Vesalius in the adult human crania in north India. For this purpose, 200 macerated skulls of unknown age and sex were observed. These skulls were obtained from the departments of Anatomy in

Direct numerical simulation (DNS) of flow past a square cylinder at a Reynolds number of 100 has been carried out to explore the effect of blowing in the form of jet(s) on vortex shedding. Higher order spatial as well as temporal discretization has been employed for the discretization of governing equations. The varying number of jets, jet velocity profiles and different blowing velocities are studied to investigate the characteristics of vortex shedding. The parabolic velocity profile has been found to be more effective in suppressing the vortex shedding as compared to the uniform velocity. Complete suppression of vortex shedding along with remarkable reduction in drag coefficient has been achieved for both jet velocity profiles but at different velocities. The corresponding values for uniform and parabolic jet profiles are 0.87 and 0.6, respectively at a mass flux of 0.120. The study also reveals that there is considerable effect of the number of jets on the vortex shedding phenomena.

No abstract

Current explanation techniques towards a transparent Convolutional Neural Network (CNN) mainly focuses on building connections between the human-understandable input features with models' prediction, overlooking an alternative representation of the input, the frequency components decomposition. In this work, we present an analysis of the connection between the distribution of frequency components in the input dataset and the reasoning process the model learns from the data. We further provide quantification analysis about the contribution of different frequency components toward the model's prediction. We show that the vulnerability of the model against tiny distortions is a result of the model is relying on the high-frequency features, the target features of the adversarial (black and white-box) attackers, to make the prediction. We further show that if the model develops stronger association between the low-frequency component with true labels, the model is more robust, which is the explanation of why adversarially trained models are more robust against tiny distortions.

No abstract