Search for: All records

Influenza viruses are a major global health burden with up to 650,000 associated deaths annually. Beyond seasonal illness, influenza A viruses (IAVs) pose a constant pandemic threat due to novel emergent viruses that have evolved the ability to jump from their natural avian hosts to humans. Because of this threat, active surveillance of circulating IAV strains in wild and domestic bird populations is vital to our pandemic preparedness and response strategies. Here, we report on IAV surveillance data collected from 2017 to 2022 from wild and domestic birds in Bangladesh. We note evidence to suggest that male birds show a higher risk of IAV, including highly pathogenic avian influenza (HPAI) A(H5) virus, positivity than female birds. The data was stratified to control for selection bias and confounding variables to test the hypothesis that male birds are at a higher risk of IAV positivity relative to female birds. The association of IAV and A(H5) largely held in each stratum, and double stratification suggested that the phenomena was largely specific to ducks. Finally, we show that chickens, male birds, and juvenile birds generally have higher viral loads compared to their counterparts. These observations warrant further validation through active surveillance across various populations. Such efforts could significantly contribute to the enhancement of pandemic prediction and risk assessment models. 

There has been a rise in automated technologies to screen potential job applicants through affective signals captured from video-based interviews. These tools can make the interview process scalable and objective, but they often provide little to no information of how the machine learning model is making crucial decisions that impacts the livelihood of thousands of people. We built an ensemble model – by combining Multiple-Instance-Learning and Language-Modeling based models – that can predict whether an interviewee should be hired or not. Using both model-specific and model-agnostic interpretation techniques, we can decipher the most informative time-segments and features driving the model's decision making. Our analysis also shows that our models are significantly impacted by the beginning and ending portions of the video. Our model achieves 75.3% accuracy in predicting whether an interviewee should be hired on the ETS Job Interview dataset. Our approach can be extended to interpret other video-based affective computing tasks like analyzing sentiment, measuring credibility, or coaching individuals to collaborate more effectively in a team. 

The flow through the collector of a solar chimney power plant model on the roof of the Aerospace and Mechanical Engineering building at the University of Arizona was investigated numerically for the conditions of the experiment. The measured wall temperature and inflow velocity for a representative day were chosen for the simulation. The simulation was performed with a newly developed higher-order-accurate compact finite difference code. The code employs fifth-order-accurate biased compact finite differences for the convective terms and fourth-order-accurate central compact finite differences for the viscous terms. A fourth-order-accurate Runge-Kutta method was employed for time integration. Unsteady random disturbances are introduced at the inflow boundary and the downstream evolution of the resulting waves was investigated based on the Fourier transforms of the unsteady flow data. Steady azimuthal waves with a wavenumber of roughly four based on the channel half-height are the most amplified as a result of Rayleigh-Benard-Poiseuille instability. Different from plane Rayleigh-Benard-Poiseuille flow, these waves appear to merge in the streamwise direction. Oblique waves are also amplified. The growth rates are however lower than for the steady modes. Because of the strong streamwise flow acceleration, the growth rates decrease in the downstream direction. 

Inward radial Rayleigh-Be'nard-Poiseuille flow can exhibit a buoyancy-driven instability when the Rayleigh number exceeds a critical value. Furthermore, similar to plane Rayleigh-Be'nard-Poiseuille flow, a viscous Tollmien-Schlichting instability can occur when the Reynolds number is high enough. Direct numerical simulations were carried out with a compressible Navier-Stokes code in cylindrical coordinates to investigate the spatial stability of the inward radial flow inside the collector of a hypothetical solar chimney power plant. The convective terms were discretized with fifth-order-accurate upwind-biased compact finite-differences and the viscous terms were discretized with fourth-order-accurate compact finite differences. For cases with buoyancy-driven instability, steady three-dimensional waves are strongly amplified. The spatial growth rates vary significantly in the radial direction and lower azimuthal mode numbers are amplified closer to the outflow. Traveling oblique modes are amplified as well. The growth rates of the oblique modes decrease with increasing frequency. In addition to the purely radial flow, a spiral flow with swept inflow was examined. Overall lower growth rates are observed for the spiral flow compared to the radial flow. Different from the radial flow, the relative wave angles and growth rates of the left and right traveling oblique modes are not identical. A plane RBP case with viscosity-driven instability by Chung et al. was considered as well. The reported growth rate and phase speed were matched with good accuracy. 

For Rayleigh-Bénard-Poiseuille flows, thermal stratification resulting from a wall-normal temperature gradient together with an opposing gravitational field can lead to buoyancy-driven instability. Moreover, for sufficiently large Reynolds numbers, viscosity-driven instability can occur. Two higher-order-accurate methods based on the full and linearized Navier-Stokes equations were developed for investigating the temporal stability of such flows. The new methods employ a spectral discretization in the homogeneous directions. In the wall-normal direction, the convective and viscous terms are discretized with fifth-order-accurate biased and fourth-order-accurate central compact finite differences. A fourth-order-accurate explicit Runge-Kutta method is employed for time integration. To validate the methods, the primary instability was investigated for different combinations of the Reynolds and Rayleigh number. The results from these primary stability investigations are consistent with linear stability theory results from the literature with respect to both the onset of the instability and the dependence of the temporal growth rate on the wave angle. For the cases with buoyancy-driven instability, strong linear growth is observed for a broad range of spanwise wavenumbers. The largest growth rates are obtained for a wave angle of 90deg. For the cases with viscosity-driven instability, the linear growth rates are lower and the first mode to experience nonlinear growth is a higher harmonic with half the wavelength of the fundamental. 

This paper reports on an experimental investigation of the flow through the collector of a solar chimney power plant which has been constructed on the roof of the Aerospace and Mechanical Engineering building at the University of Arizona. This model contains a central chimney which is a long tubular structure located in the center and a circular collector that employs the greenhouse effect to heat up the air under it. The chimney is 5.9m high and the collector radius measured from the center of the chimney is 4.13m. Measurements were carried out from April to June 2019. Several types of J thermocouples were mounted inside the collector at various radial locations to measure the air temperature both near the ground and the ceiling of the collector. A hot-wire probe (anemometer) was employed to measure the airflow velocity under the collector near the chimney inlet. A traverse system was designed and constructed, which allows the anemometer to be moved in the radial and circumferential direction under the collector. The height of the probe position above the collector ground can be adjusted by rotating the probe support along its longitudinal axis. A digital analog conversion system was used to convert the thermocouple and hot-wire readouts into binary data for processing via a LabVIEW interface. In parallel to the experiments, high-fidelity numerical simulations are being carried out for the conditions of the experiments. The simulations show longitudinal flow structures near the collector outflow that likely result from a buoyancy-driven instability. 

For plane channel flow, thermal stratification resulting from a wall-normal temperature gradient together with an opposing gravitational field can lead to buoyancy-driven instability of three-dimensional waves. Moreover, viscosity-driven instability can lead to the amplification of two-dimensional Tollmien-Schlichting waves. Temporal stability simulations considering different combinations of Reynolds number and Rayleigh number were performed to investigate both the buoyancy and viscosity-driven instability of Rayleigh-Benard-Poiseuille flow. The investigated cases are either (1) stable, (2) unstable with respect to three-dimensional waves (buoyancy-driven instability), or (3) unstable with respect to two-dimensional waves (viscosity-driven instability). Two new and highly accurate computational fluid dynamics codes have been developed for solving the full and linearized unsteady compressible Navier-Stokes equations in Cartesian coordinates. The codes employ fifth-order-accurate upwind-biased compact finite differences for the convective terms and fourth-order-accurate compact finite differences for the viscous terms. For the case with buoyancy-driven instability, strong linear growth is observed for a broad range of spanwise wavenumbers and the wavelength of the spanwise mode with the strongest non-linear growth is gradually decreasing in time. For the case with viscosity-driven instability, the linear growth rates are lower and the first mode to experience non-linear growth is a higher harmonic with half the wavelength of the primary wave. The present results are consistent with the neutral curves from the linear stability theory analysis by Gage and Reid. 

Recent Transformer-based contextual word representations, including BERT and XLNet, have shown state-of-the-art performance in multiple disciplines within NLP. Fine-tuning the trained contextual models on task-specific datasets has been the key to achieving superior performance downstream. While fine-tuning these pre-trained models is straightforward for lexical applications (applications with only language modality), it is not trivial for multimodal language (a growing area in NLP focused on modeling face-to-face communication). More specifically, this is due to the fact that pre-trained models don’t have the necessary components to accept two extra modalities of vision and acoustic. In this paper, we proposed an attachment to BERT and XLNet called Multimodal Adaptation Gate (MAG). MAG allows BERT and XLNet to accept multimodal nonverbal data during fine-tuning. It does so by generating a shift to internal representation of BERT and XLNet; a shift that is conditioned on the visual and acoustic modalities. In our experiments, we study the commonly used CMU-MOSI and CMU-MOSEI datasets for multimodal sentiment analysis. Fine-tuning MAG-BERT and MAG-XLNet significantly boosts the sentiment analysis performance over previous baselines as well as language-only fine-tuning of BERT and XLNet. On the CMU-MOSI dataset, MAG-XLNet achieves human-level multimodal sentiment analysis performance for the first time in the NLP community. 

Since the generation of green and clean renewable energy is a major concern in the modern era, solar energy conversion technologies such as the solar chimney power plant are gaining more attention. In order to accurately predict the performance of these power plants, hydrodynamic instabilities that can lead to large-scale coherent structures which affect the mean flow, have to be identified and their onset has to be predicted accurately. The thermal stratification of the collector flow (resulting from the temperature difference between the heated bottom surface and the cooled top surface) together with the opposing gravity can lead to buoyancy-driven instability. As the flow accelerates inside the collector, the Reynolds number can get large enough for viscous (Tollmien-Schlichting) instability to occur. A new highly accurate compact finite difference Navier-Stokes code in cylindrical coordinates has been developed for the spatial stability analysis of such radial flows. The new code was validated for a square channel flow. Stability results for different stable and unstable Reynolds/Rayleigh number combinations were in good agreement with temporal stability simulations as well as linear stability theory. To investigate the radial flow effect, spatial stability simulations were then carried out for a computational domain with constant streamwise extent and different outflow radii. The Reynolds and Rayleigh number were chosen such that buoyancy-driven instability occured. For cases with significant radial flow effect, the spatial growth rates of the azimuthal modes were found to vary considerably in the streamwise direction.