I utilised household data from Cambodia, obtained from the Demographic and Health Surveys (DHS) Program, in order to classify households by poverty level. I implemented Softmax Classification, Random Forest, and Neural Network Models for poverty classification. Finally, I identified important features in distinguishing households by wealth value.

The flanged parallel-plate waveguide is analyzed based on the method of Kobayashi potentials (KP) using Fourier functional space. The presentation of the method is free from intricate mathematics. Standard integral identities are used for problem formulation, without direct usage of Weber-Schafheitlin (WS) integrals. The Fourier function space is exploited for the construction of the governing linear equations instead of Jacobi polynomials. A simple strategy is suggested for the evaluation of the required improper integrals. Radiation into lossless and lossy media is considered. Near-field results are rigorously validated through convergence analysis. Far-field patterns are compared with predictions of the surface equivalence theorem (SET).

A large and growing volume of repeat Digital Elevation Models (DEMs) obtained from multiple spaceborne sensors enables high-resolution mapping of the Earth’s surface at continental scales. Presented is a method for efficiently mosaicking sets of repeat, overlapping DEMs using their pairwise, translational offsets to remove poor quality DEMs and optimize their alignment prior to merging. The Coregistration, Adjustment and Median of Stacks (CAMS) approach provides substantially higher quality and accuracy than the median of overlaps when compared to high resolution LiDAR data. The method is general and adaptable for large-area mosaic production

Medical image segmentation is essential in medical image analysis since it can provide reliable assistance in computer-aided clinical diagnosis, treatment planning, and intervention. Although deep learning algorithms based on CNNs and Transformers have made notable progress in medical image segmentation, it is still challenging owing to the objects with complex structures, low discrimination and differences between individuals. To alleviate the problems, we propose a novel 3D medical image segmentation network based on Transformers and CNNs combining morphological information and reticular mechanism. Firstly, the morphological constraint stream is designed to learn the prior shape information based on the CNN model for enhancing the interpretability of the ultimate trained model and accelerating the convergence. Secondly, the Reticular Transformer is utilized to obtain multi-scale information based on the Transformer, which can bind the local texture information and underlying semantic information to further acquire the feature maps with sufficient details and receptive field. The experiments demonstrate that our proposed method outperforms many existing segmentation models in terms of the performance in metrics DSC and HD (80.46\% in DSC on the Synapse dataset and 90.83\% in DSC on the ACDC dataset). The code will be released at https://github.com/rocklijun/MCRformer. Our proposed method can not only achieve superior performance compared with most of the current state-of-the-art methods, but also enhance the robustness and interpretability of the model. Furthermore, the proposed morphological constraint stream has the potential to be transferred to other frameworks for different medical image analysis tasks.

The widespread adoption of Artificial Intelligence (AI) models by various industries in recent years have made Explainable Artificial Intelligence (XAI) an active field of research. This adoption can cause trust and effectiveness to suffer if the results of these models are not favorable in some way. XAI has advanced to the point where many metrics have been proposed as reasons for the outputs of many AI models. However, there is little consensus about what technical metrics are most important, nor is there a consensus on how best to analyze explainable methods and models. A discussion of varying attempts at this is brought forth, but the paper also goes into the ethics of AI and its societal impact. Given the modern ubiquity with which AI exists and the immensely multidisciplinary approach, AI has evolved into, using only technical metrics cannot fully describe XAI’s effectiveness. This paper explores several approaches to measuring the ethical effects of XAI, whether it has any bearing on modern research, as well as how the impacts of AI and XAI are measured on society. The full attempt at quantifying XAI models’ effectiveness is explored from a technical and non-technical point of view.

How cost-efficient are potential infrastructure sharing business models for the 5G/6G era? How should telecommunications regulators formulate a national policy to improve mobile broadband connectivity? These are significant questions that need to be addressed if we are to deliver universal, affordable broadband and achieve Target 9c of the UN Sustainable Development Goals.

Terahertz attenuated total reflectance (THz-ATR) is being extensively used for spectroscopy and imaging applications in materials science, chemistry, and biomedical research. One of the driving motivations is its potential for studying protein dynamics, identifying living tissues, or detecting herbicide traces, as a non-destructive and non-ionizing technique. It is well known that studying biological and chemical samples in the terahertz range represents a real challenge because of the presence of water, which is highly absorbing. In this work, we present experimental data from THz-ATR spectroscopy of water in the 1-15 THz frequency range and propose an insightful interpretation of the optical phenomenon that is taking place in this experiment, supported by finite-element-method simulations and an analytical model. Interestingly, this study reveals that there is no evanescent field at the interface between the crystal and water, but that a refracted beam is propagating through the water, independently from the value of the angle of incidence. These findings improved the signal intensity after a simple modification of the accessory, by creating a thin film of micrometer thickness with a top mirror. We believe that this technique could lead to highly sensitive terahertz microfluidic systems for chemical and biomolecular detection.

Enhanced supercontinuum generation (SCG) is experimentally demonstrated in integrated silicon nitride (Si3N4) waveguides incorporating highly nonlinear graphene oxide (GO) in the form of two-dimensional (2D) films. On-chip integration of the 2D GO films with precise control of their thickness is realized by using a transfer-free and layer-by-layer coating method. The control of the film length and coating position is achieved via window opening in the upper silica cladding of the photonic integrated chips. Detailed SCG measurements are performed using the fabricated devices with different waveguide geometries and GO film thicknesses, and the results are compared with devices without GO. Significantly improved spectral broadening of ultrashort optical pulses with ultrahigh peaks powers exceeding 1000 W is observed for the hybrid devices, achieving up to 2.4 times improvement in the spectral bandwidth relative to devices without GO. Theoretical analyses for the influence of GO film thickness, coating length, coating position, and waveguide geometry are also provided by fitting the experimental results with theory, showing that there is still significant room for further improvement. This work opens up a promising new avenue towards improving the SCG performance of photonic integrated devices by incorporating functional 2D materials.

Despite decades of research, the automatic detection of vulnerabilities in mobile apps remains an open challenge. Among the possible solutions, SAST tools uncover source or compiled code security flaws without needing the app to be executed and tested in a controlled environment. However, SAST tools share several limitations, such as the detection of narrowed vulnerability classes, lack of updates, and limited resiliency to obfuscation techniques. This paper presents SEBASTiAn, a black-box automatic static analysis tool for security vetting Android and iOS apps. It relies on a modular approach to cope with new vulnerabilities.

Detection of road boundaries using camera-LIDAR fusion in world coordinate system poses several challenges. First, the pixels from the image need to be transformed into the world coordinate system, which is commonly done using inverse perspective mapping (IPM) technique. However, this technique is built based on assumptions that the road model is planar and rigid, assumptions that are not always held in real-world scenarios. Furthermore, obtaining the ground truth road boundaries is a difficult task for a real-world test, especially when the detection results are represented in world coordinate system. To solve the first challenge, this paper proposes a new framework for fusing camera and LIDAR data for detecting road boundaries using a transformation technique that can work on scenarios with varying and nonplanar roads. With this framework, we were able to fuse road boundary detection results that are extracted from camera and LIDAR separately. To overcome the second challenge, we used Carla Simulator and developed a method for extracting the ground truth points so that evaluation and comparison of various road boundary detection methods can be done properly. Based on our test results, the proposed transformation technique performed better than the widely used IPM technique. The results also show that our road boundary detection method can achieve error values below 0.3 m and reveal some key characteristics of camera and LIDAR-based detection method. Those characteristics were utilized to improve the fusion technique that is demonstrated in this paper.

Time-harmonic inverse source solutions are commonly working with electric and magnetic surface current densities defined and discretized on appropriately chosen Huygens surfaces. An efficient meshless alternative are spherical harmonics expansions of low order distributed along the chosen Huygens surface, which still possess pretty good spatial localization properties. Similar to expansions with electric and magnetic surface current densities, distributed expansions with Cartesian spherical harmonics or standard vector spherical harmonics are redundant when they are placed on Huygens surfaces. This redundancy is reduced by allowing only spherical harmonics, which can be excited by surface current density distributions in a specific plane, and by forming directive spherical harmonics. This results in a considerable reduction of the necessary number of unknowns for representing the sources and improved conditioning of the discretized integral equations. Inspired by the Huygens radiator concept, appropriate directive harmonics on the basis of Cartesian and vector spherical harmonics are derived, implemented, and compared in terms of their performances. The validity of the presented techniques is confirmed via inverse source solutions for synthetic and real measurement data.

Amid the current digital transformation wave, predictive maintenance (PdM) using machine learning has become prevalent due to its potential to reduction for total cost and turnaround time in the manufacturing and MRO (maintenance, repair, and operations) sectors. Predicting machinery tool wear or remaining useful life is one of the essential applications in modern PdM. As a result, many deep learning based prediction methods emerged in recent years’ literature and continuously advancing the state-of-the-art. In this paper, we propose a tool wear prediction framework that orchestrates a range of machine learning techniques to enhance prediction accuracy. These techniques include sequence modeling with convolutional neural networks, synthetic data generation with conditional generative adversarial networks, and customized regularization strategy. Experiments were performed with two public milling datasets to evaluate the effectiveness of the proposed framework. Results show that our prediction model, sCNN-Ex, outperforms state-of-the-art tool wear prediction methods.

In this study, we aimed at investigating generalizability of GFCE-MRI model using data from seven institutions by manipulating heterogeneity of training MRI data under two popular normalization approaches. A multimodality-guided synergistic neural network (MMgSN-Net) was applied to map from T1-weighted and T2-weighted MRI to contrast-enhanced MRI (CE-MRI) for GFCE-MRI synthesis in patients with nasopharyngeal carcinoma. MRI data from three institutions were used separately to generate three uni-institution models and jointly for a tri-institution model. Min-Max and Z-Score were applied for data normalization of each model. MRI data from the remaining four institutions served as external cohorts for model generalizability assessment. Quality of GFCE-MRI was quantitatively evaluated against ground-truth CE-MRI using mean absolute error (MAE) and peak signal-to-noise ratio (PSNR). Results showed that performance of all uni-institution models remarkably dropped on the external cohorts. By contrast, model trained using multi-institutional data with Z-Score normalization yielded significantly improved model generalizability

This paper aims to clarify the impact of channel aging and phase noise on the performance of intelligent reflecting surface-aided wireless systems. We first model mathematically the outdated channel state information (CSI) due to Doppler shifts and phase noise stemming from hardware impairment. Then, a closed-form expression of achievable spectral efficiency under noisy and aged CSI is theoretically derived. Some typical simulation results to numerically demonstrate the performance impact are illustrated.

Background Accurate needle placement is crucial during minimally invasive procedures such as biopsies, regional anesthesia, and fluid aspiration. 2D Ultrasound is widely used for needle guidance during such procedures, however, it has a limited field-of-view and poor needle visibility for steep insertion angles. Methods In this work, we propose a novel machine learning (ML)-based method for real-time needle segmentation in handheld 2D ultrasound systems. The proposed method involves a fast and simple annotation technique allowing for the labeling of large datasets. It then utilizes the U-Net architecture which is modified to allow for easy integration into a handheld ultrasound system. Two datasets were used in this work, one consisting of B-mode ultrasound videos obtained from human tissue and the other consisting of videos and frames from chicken, porcine and bovine tissue. The model is trained on 1262 frames and evaluated on 209 frames. Results This approach achieves an Intersection Over Union (IoU) of 0.75 and a dice coefficient of 0.851 on frames obtained from human tissue. The model is integrated into the processing pipeline of a portable ultrasound system and achieves an overall processing speed of about 8 frames per second. The proposed approach outperforms state-of-the-art methods for needle segmentation while achieving real-time integration. This work is a step forward towards real-time needle guidance using machine learning-based algorithms in handheld ultrasound systems.

Our study considered the problem of classifying routine clinical EEG. We incorporated NLP tools into the workflow for time series classification. We transformed EEG signals into strings of symbols. We then applied byte-pair encoding (BPE) to split the new text into combinations of symbols or tokens, each associated with different patterns of changes in EEG amplitude. We validated the proposed workflow under the framework of predicting patients’ biological age.

Facial expression recognition (FER) is a challenging job in Computer Vision due to data uncertainties rooted in the ambiguity of facial expressions. As a complement to current FER studies huddling in data-level or feature-level for suppressing such uncertainties,  we propose a simple yet efficient coarse-to-fine learning strategy at task-level inspired by human beings’ emotion cognitive mode. Specifically, a child learns quickly whether his behavior is allowed by reading adults’ facial expressions for coarse attitude like positive or negative, and then adjusts accordingly via further interpreting its fine sentiment meaning like happy or angry. .

*This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. In this article, we propose a novel image-based method for radar cross-section (RCS) prediction of a cluster of multiple static targets by synthesizing replicas of an original radar image. In this approach, we first measure the near-field backscattering of a target and reconstruct a corresponding radar image. Then, modified copies of this image with rotation, translation, and spatial filtering, are generated according to the predefined desired arrangement, and they are coherently summed to create a single synthesized image in which all scattering contributions contained in the modified images are virtually included. Finally, the synthesized image is utilized to predict the far-field RCS of the multiple targets, based on the theory of image-based near-field-to-far-field transformation (NFFFT). By employing the proposed algorithm, we can avoid building multiple test targets, resulting in the reduction of the production costs of them. Moreover, we can easily test several different experimental layouts of the multiple targets without repeating a real measurement. Numerical simulations and experiments are provided to demonstrate the validity of the proposed image-based RCS synthesis.