A key solution to visual question answering (VQA) exists in how to fuse visual and language features extracted from an input image and question. We show that an attention mechanism that enables ...dense, bi-directional interactions between the two modalities contributes to boost accuracy of prediction of answers. Specifically, we present a simple architecture that is fully symmetric between visual and language representations, in which each question word attends on image regions and each image region attends on question words. It can be stacked to form a hierarchy for multi-step interactions between an image-question pair. We show through experiments that the proposed architecture achieves a new state-of-the-art on VQA and VQA 2.0 despite its small size. We also present qualitative evaluation, demonstrating how the proposed attention mechanism can generate reasonable attention maps on images and questions, which leads to the correct answer prediction.
This paper presents a comprehensive review on the development of higher-order continuum models for capturing size effects in small-scale structures. The review mainly focus on the size-dependent ...beam, plate and shell models developed based on the nonlocal elasticity theory, modified couple stress theory and strain gradient theory due to their common use in predicting the global behaviour of small-scale structures. In each higher-order continuum theory, various size-dependent models based on the classical theory, first-order shear deformation theory and higher-order shear deformation theory were reviewed and discussed. In addition, the development of finite element solutions for size-dependent analysis of beams and plates was also highlighted. Finally a summary and recommendations for future research are presented. It is hoped that this review paper will provide current knowledge on the development of higher-order continuum models and inspire further applications of these models in predicting the behaviour of micro- and nano-structures.
A BCMO-ANN algorithm for vibration and buckling optimization of functionally graded porous (FGP) microplates is proposed in this paper. The theory is based on a unified framework of higher-order ...shear deformation theory and modified couple stress theory. A combination of artificial neural network (ANN) and balancing composite motion optimization (BCMO) is developed to solve the optimization problems and predict stochastic vibration and buckling behaviors of functionally graded porous microplates with uncertainties of material properties. The characteristic equations are derived from Hamilton’s principle and approximation of field variables under Ritz-type exponential series. Numerical results are obtained to investigate the effects of the material distribution, material length scale, porosity density and boundary conditions on natural frequencies and critical buckling loads of functionally graded porous microplates. The novel results derived from this paper can be used as future references.
•A BCMO-ANN algorithm for stochastic vibration and buckling analysis of functionally graded porous microplates is presented.•A unified framework of higher-order shear deformation theory and modified couple stress theory is presented.•Ritz method with hybrid shape functions yields fast convergence and accurate results.•Effects of materials and geometry on stochastic behaviors of functionally graded porous microplates are investigated.
A MIL-53(Fe) analogue was successfully synthesized by a HF free-solvothermal method. The sample was characterized by XRD, N
2
adsorption (BET), TEM, FTIR, XPS and AAS. From the N
2
...adsorption-desorption isotherms, it can be seen that the structure of MIL-53(Fe) in the anhydrous form exhibits closed pores with almost no accessible porosity to nitrogen gas. The XPS results reveal that Fe is really incorporated into the MIL-53(Fe) framework. In the hydrated form, the pores of MIL-53(Fe) are filled with water molecules. Thus, MIL-53(Fe) exhibited a very high adsorption capacity of As(
v
) in aqueous solution (
Q
max
of 21.27 mg g
−1
). Adsorption kinetics data revealed that As(
v
) adsorption isotherms fit the Langmuir model and obey the pseudo-second-order kinetic equation.
A MIL-53(Fe) analogue was successfully synthesized by a HF free-solvothermal method.
Iris recognition refers to the automated process of recognizing individuals based on their iris patterns. The seemingly stochastic nature of the iris stroma makes it a distinctive cue for biometric ...recognition. The textural nuances of an individual's iris pattern can be effectively extracted and encoded by projecting them onto Gabor wavelets and transforming the ensuing phasor response into a binary code - a technique pioneered by Daugman. This textural descriptor has been observed to be a robust feature descriptor with very low false match rates and low computational complexity. However, recent advancements in deep learning and computer vision indicate that generic descriptors extracted using convolutional neural networks (CNNs) are able to represent complex image characteristics. Given the superior performance of CNNs on the ImageNet large scale visual recognition challenge and a large number of other computer vision tasks, in this paper, we explore the performance of state-of-the-art pre-trained CNNs on iris recognition. We show that the off-the-shelf CNN features, while originally trained for classifying generic objects, are also extremely good at representing iris images, effectively extracting discriminative visual features and achieving promising recognition results on two iris datasets: ND-CrossSensor-2013 and CASIA-Iris-Thousand. We also discuss the challenges and future research directions in leveraging deep learning methods for the problem of iris recognition.
The coronavirus disease 2019 (COVID-19) epidemic affects people's health and health-related quality of life (HRQoL), especially in those who have suspected COVID-19 symptoms (S-COVID-19-S). We ...examined the effect of modifications of health literacy (HL) on depression and HRQoL. A cross-sectional study was conducted from 14 February to 2 March 2020. 3947 participants were recruited from outpatient departments of nine hospitals and health centers across Vietnam. The interviews were conducted using printed questionnaires including participants' characteristics, clinical parameters, health behaviors, HL, depression, and HRQoL. People with S-COVID-19-S had a higher depression likelihood (OR, 2.88;
< 0.001), lower HRQoL-score (B, -7.92;
< 0.001). In comparison to people without S-COVID-19-S and low HL, those with S-COVID-19-S and low HL had 9.70 times higher depression likelihood (
< 0.001), 20.62 lower HRQoL-score (
< 0.001), for the people without S-COVID-19-S, 1 score increment of HL resulted in 5% lower depression likelihood (
< 0.001) and 0.45 higher HRQoL-score (
< 0.001), while for those people with S-COVID-19-S, 1 score increment of HL resulted in a 4% lower depression likelihood (
= 0.004) and 0.43 higher HRQoL-score (
< 0.001). People with S-COVID-19-S had a higher depression likelihood and lower HRQoL than those without. HL shows a protective effect on depression and HRQoL during the epidemic.
This paper presents a simple two-variable shear deformation theory for bucking, bending, and vibration behaviours of functionally graded porous (FGP) beams. The displacement field of beams is ...developed from the separation of variables. Three typical porosity distribution types are considered. Mass density and elastic modulus of FGP beams are assumed to be graded in the beam’s thickness. Governing equations are derived from Lagrange’s principle. The exponential approximation functions are developed for various boundary conditions to predict frequency, buckling load, deflection, and stress of beams. The effects of boundary condition, span-to-height ratio, porous distribution pattern, porosity parameter, and shear deformation on the critical buckling load, frequency, deflection, and stress of beams are investigated in detail.
The COVID-19 pandemic has imposed a heavy burden on health care systems and governments. Health literacy (HL) and eHealth literacy (as measured by the eHealth Literacy Scale eHEALS) are recognized as ...strategic public health elements but they have been underestimated during the pandemic. HL, eHEALS score, practices, lifestyles, and the health status of health care workers (HCWs) play crucial roles in containing the COVID-19 pandemic.
The aim of this study is to evaluate the psychometric properties of the eHEALS and examine associations of HL and eHEALS scores with adherence to infection prevention and control (IPC) procedures, lifestyle changes, and suspected COVID-19 symptoms among HCWs during lockdown.
We conducted an online survey of 5209 HCWs from 15 hospitals and health centers across Vietnam from April 6 to April 19, 2020. Participants answered questions related to sociodemographics, HL, eHEALS, adherence to IPC procedures, behavior changes in eating, smoking, drinking, and physical activity, and suspected COVID-19 symptoms. Principal component analysis, correlation analysis, and bivariate and multivariate linear and logistic regression models were used to validate the eHEALS and examine associations.
The eHEALS had a satisfactory construct validity with 8 items highly loaded on one component, with factor loadings ranked from 0.78 to 0.92 explaining 76.34% of variance; satisfactory criterion validity as correlated with HL (ρ=0.42); satisfactory convergent validity with high item-scale correlations (ρ=0.80-0.84); and high internal consistency (Cronbach α=.95). HL and eHEALS scores were significantly higher in men (unstandardized coefficient B=1.01, 95% CI 0.57-1.45, P<.001; B=0.72, 95% CI 0.43-1.00, P<.001), those with a better ability to pay for medication (B=1.65, 95% CI 1.25-2.05, P<.001; B=0.60, 95% CI 0.34-0.86, P<.001), doctors (B=1.29, 95% CI 0.73-1.84, P<.001; B 0.56, 95% CI 0.20-0.93, P=.003), and those with epidemic containment experience (B=1.96, 95% CI 1.56-2.37, P<.001; B=0.64, 95% CI 0.38-0.91, P<.001), as compared to their counterparts, respectively. HCWs with higher HL or eHEALS scores had better adherence to IPC procedures (B=0.13, 95% CI 0.10-0.15, P<.001; B=0.22, 95% CI 0.19-0.26, P<.001), had a higher likelihood of healthy eating (odds ratio OR 1.04, 95% CI 1.01-1.06, P=.001; OR 1.04, 95% CI 1.02-1.07, P=.002), were more physically active (OR 1.03, 95% CI 1.02-1.03, P<.001; OR 1.04, 95% CI 1.03-1.05, P<.001), and had a lower likelihood of suspected COVID-19 symptoms (OR 0.97, 95% CI 0.96-0.98, P<.001; OR 0.96, 95% CI 0.95-0.98, P<.001), respectively.
The eHEALS is a valid and reliable survey tool. Gender, ability to pay for medication, profession, and epidemic containment experience were independent predictors of HL and eHEALS scores. HCWs with higher HL or eHEALS scores had better adherence to IPC procedures, healthier lifestyles, and a lower likelihood of suspected COVID-19 symptoms. Efforts to improve HCWs' HL and eHEALS scores can help to contain the COVID-19 pandemic and minimize its consequences.
•Introduce a novel approach utilizing C3D for extracting spatio-temporal features.•Develop a novel feature-level fusion relying on multimodal compact bilinear pooling.•Demonstrate significant ...potential to detect emotions from spatio-temporal information.•Show our system’s adaptability to sufficiently combine different types of multimodal streams (i.e. audio and video, face and body).•Develop the most efficient multimodal emotion recognition systems on both FABO and eNTERFACE dataset which are more applicable for real-time implementation.
Multimodal emotion recognition has attracted great interest recently and numerous methodologies have been successfully investigated. However, the task requires the effective fusion multimodal representations in audio and video domains, and existing approaches still perform poorly on such a challenging task. This paper proposes a novel framework for recognizing emotion from multiple sources including facial expression, pose, body movements, and voice. In this framework, we first introduce new deep spatio-temporal features by cascading 3-dimensional convolution neural networks (C3Ds) and deep belief networks (DBNs) to effectively model spatial and temporal information presented in video and audio for emotion recognition. We subsequently propose a new feature-level fusion approach based on a bilinear pooling theory to combine the visual and audio feature vectors. The proposed fusion strategy allows all elements of the component vectors to interact with each other in an effective way, resulting in expressively capturing the complex and intrinsic associations between the component modalities. Extensive experiments conducted on the eNTERFACE and FABO multimodal emotion databases demonstrate that our proposed system leads to improved multimodal emotion recognition performance and significantly outperforms recent state-of-the-art approaches.
This paper presents a Ritz-type analytical solution for buckling and free vibration analysis of functionally graded (FG) sandwich beams with various boundary conditions using a quasi-3D beam theory. ...It accounts a hyperbolic distribution of both axial and transverse displacements. Equations of motion are derived from Lagrange’s equations. Two types of FG sandwich beams namely FG-faces ceramic-core (type A) and FG-core homogeneous-faces (type B) are considered. Numerical results are compared with earlier works and investigated effects of the power-law index, thickness ratio of layers, span-to-depth ratio and boundary conditions on the critical buckling loads and natural frequencies.