Given the recent advancements in deep learning (DL), and computing capabilities, there has been significant progress in the development of systems analyzing facial behavior, evolving from images under controlled laboratory conditions, to more challenging real-world scenarios, including video-based recognition. However, DL models typically require training with large-scale datasets to provide a high level of performance. The collection and annotation of such datasets is a costly undertaking that relies on domain experts. Moreover, the annotation process is highly vulnerable to the ambiguity of expressions or Action Units (AUs) due to the bias induced by experts. This paper provides a comprehensive review of state-of-the-art weakly supervised learning (WSL) methods for facial behavior analysis based on images and videos. First, we introduce a taxonomy of relevant WSL scenarios, and then methods proposed for classification applications (using weak categorical/discrete labels), and for regression applications (using weak ordinal/dimensional labels). For both types of recognition applications (classification and regression), we provide a systematic review of state-of-the-art machine learning (ML) models for expression and AU recognition, and intensity estimation in each WSL scenario, discussing their strengths and limitations. An overview of widely used experimental methodologies (public datasets and protocols) is also provided for the performance evaluation of these state-of-the-art models. Finally, our critical analysis of the methods, along with their experimental results, provides insight into the key challenges of WSL in different scenarios, motivating future research to leverage weakly labeled data for real-world facial behavior analysis problems. Our review indicates that WSL methods may provide a cost-effective approach to train robust ML/DL models for facial behavior analysis in real-world scenarios. A website including constantlyupdated survey is provided at https://github.com/praveena2j/ awesome-Weakly-Supervised-Facial-Behavior-Analysis.

Modern academic research encompasses a broad spectrum of subjects, reflecting the intricate and evolving challenges and opportunities that societies encounter globally. This study delves into the role of literature in shaping cultural identities and the profound impact of technology on traditional structures. Current scholarly contributions offer deep insights into the blend of humanistic concerns and technological advancements, underscoring their combined relevance in tackling contemporary societal issues.

Passive synthetic aperture radar (SAR) has attracted research in recent years because of its advantages in exploiting opportunistic signals for imaging. Hence, it minimizes the need for active radar transmitters, significantly reducing system costs. In addition, the massive proliferation of telecommunication satellites in recent years motivates the idea of using the signal for the integrated sensing and communication (ISAC) concept. This paper explores the feasibility and opportunity of exploiting communication signals, specifically the common waveform of orthogonal frequency division multiplexing (OFDM), as a source of SAR illumination. The reflected signal is captured using a flying uncrewed aerial vehicle (UAV) at a much lower altitude than the satellite. To perform this evaluation, we developed an extensive signal-level emulator that considers many realistic aspects, such as orbital parameters, pulse repetition frequency (PRF), RF radio bandwidth, and antenna beamwidth. Point targets were placed within the imaging swath using regular spacing, and then the backscattered signal was collected using the UAV. Accordingly, the SAR image was formed using the range-Doppler (RDA) processing algorithm. The integrated sidelobe ratio (ISLR) was evaluated for ground reference point (GRP) cross-sections in order to provide a numerical assessment for performance. The results indicate the feasibility of exploring the direction of ISAC in passive SAR image formation.

In this paper, bikeshare data from Chicago on weather-friendly days in 2019 and 2020 were analyzed to investigate the variation in bikeshare travel before and during the pandemic. Our results show that bikeshare trips during the pandemic were much longer than prior to the pandemic. The increased rate of bikeshare usage was unbalanced spatially and varied significantly for different user types. Bikeshare was used remarkably more by casual users than by subscribers, and the increase occurred much more in the outskirts of the city. The increase in bikeshare travel was associated with a reduction in travel by ridehailing and public transit, especially in the urban periphery. The correlation of bikeshare use with the bus system was much less significant than with the rail system. Bike lanes/facilities had a mixed effect on bikeshare travel. Weekend bike trips increased in areas where there was no bike lane. Weekday trips, on the contrary, increased in the vicinity of bike greenways.

In this paper, we propose a novel Privacy-Preserving clearance mechanism for Local Energy Markets (PP-LEM), designed for computational efficiency and social welfare. PP-LEM incorporates a novel competitive game-theoretical clearance mechanism, modelled as a Stackelberg Game. Based on this mechanism, a privacy-preserving market model is developed using a partially homomorphic cryptosystem, allowing buyers’ reaction function calculations to be executed over encrypted data without exposing sensitive information of both buyers and sellers. The comprehensive performance evaluation demonstrates that PP-LEM is highly effective in delivering an incentive clearance mechanism with computational efficiency, enabling it to clear the market for 200 users within the order of seconds while concurrently protecting user privacy. Compared to the state of the art, PP-LEM achieves improved computational efficiency without compromising social welfare while still providing user privacy protection.

Battery Energy Storage systems play a significant role in renewable energy grids, where fault detection is critical to ensuring reliability, safety, and optimal performance. Existing methods for fault detection, which are model-based Kalman filters and data-driven signal processing techniques, could be effective but face limitations in detecting slight issues or early signs of deterioration. This paper presents a novel hybrid approach to real-time fault detection which integrates signal processing techniques with Kalman filter-based state estimation. The proposed model is designed to detect faults and predict degradation trends, thereby enhancing the overall health monitoring of battery systems. This detailed methodology covers signal acquisition, preprocessing, feature extraction, Kalman filter design, and residual analysis. Kalman filter predicts system state and performs residual analysis to identify faults. Post-Kalman filter signal processing methods like wavelet transforms are used to analyze the residuals and detect fault-specific patterns, allowing for precise fault classification, including overcharging and short circuits. This hybrid approach combines the strengths of real-time state estimation and signal processing to advance real-time battery health monitoring, which results in a robust, scalable solution for fault detection in battery Storage Systems.

The diverse range of Internet of Things (IoT) devices available today offers great potential for home automation and Smart Home (SH) development, whether partial or complete. However, for research in this area to advance, it is essential to overcome the scarcity of data, as data collection in SHs faces significant challenges, such as the complexity of devices and environments, privacy and security concerns, and restrictions imposed by proprietary devices. Furthermore, freely available datasets often have limitations, such as incomplete annotations, restricted device usage, and insufficient spatial and temporal coverage. To address this limitation, we developed a novel SH simulator that supports multi-state devices and captures detailed information on device usage and inhabitant activities, including metrics like lamp brightness and air conditioning settings (temperature, mode, etc.). The primary goal is to provide researchers in SHs, the SH, and related fields with a tool for generating datasets that effectively represent an inhabitant or group of inhabitant activity patterns in simulated environments. Home Environment Simulator Targeting Inhabitant Activities (HESTIA) resolves the inconsistency in message exchanges between devices, which often stems from differences in device manufacturers, while also eliminating the time-consuming and costly process of collecting real-world data by generating synthetic datasets. The simulator HESTIA offers rich, dynamic information that can be tailored to specific user profiles. To validate our method, we collected data over two weeks from a real environment using four devices and compared it to a synthetic version of the same setup generated by HESTIA. The results show that the simulator accurately replicates real-world environments, even for devices with inconsistent status changes. Additionally, we evaluated the computational complexity of this simulator in terms of time.

Adders are the most important element in the ALU. Speeding the addition improves the performance of the ALU in particular and the CPU in general. Pipelining is used to process tasks in parallel with minimal adjustment of the hardware needed to process the same tasks sequentially. In this paper, a pipelining procedure is presented to enable the half adder to perform the addition of two numbers in parallel. The resulted speed up factor is guaranteed to perform the same addition in half the clock cycle of full adder.

This paper introduces a novel technique based on a planar microwave sensor for contactless monitoring of hydrocarbon contamination in groundwater. Short-chain hydrocarbons are more toxic and volatile, which makes their detection vital for the management of Petroleum Hydrocarbons (PHCs) pollution in aquatic environments. The sensor consists of a tag and reader, which are wirelessly connected; hence, non-contact and highly sensitive sensing is enabled. The tag is fabricated on a flexible substrate for ease in the monitoring of the alkanes. This non-contact nature, coupled with high resolution, makes this sensor ideal for monitoring hazardous and toxic hydrocarbons in groundwater. The detection of Hexane has been studied by varying its concentration in water up to 30% (v/v).

An effective passive source localization technique is proposed as essential for industrial applications. Extensive research has been conducted on multi-station passive localization using two-dimensional angle of arrival (2D AOA), time difference of arrival (TDOA), and their combination. However, certain practical factors limit the observation to only a restricted set of one-dimensional angles of arrival (1D AOA). Additionally, there may be perturbations in the stations' locations. To address these issues, this paper proposes a localization model that hybrid a single 1D AOA with multiple TDOA measurements, taking into account the perturbations in station locations. For clarity, we refer to this combination as hybrid measurements. Two types of localization algorithms are proposed for the hybrid measurement model. The first type approximates the localization problem as a non-convex constrained weighted least squares optimization problem. We introduce two solution algorithms: the two-step weighted least squares algorithm (TWLS) and the semi-infinite relaxation method (SDR), both of which do not require an initial guess. The second type directly transforms the localization problem into a maximum likelihood estimation (MLE) minimum optimization problem. Several iterative algorithms are introduced, including iterative weighted least squares based on Taylor expansion (IRLS-TE), Gaussian Newton (GN), adaptive moment estimation (Adam), and simplification adaptive moment estimation (SAdam). Unlike the first type, these algorithms require an initial guess. Furthermore, we derive the Cramér-Rao Lower Bound (CRLB) for hybrid measurements and analyze the computational complexity and minimum station configuration needed for each solution algorithm. Simulations and the SWellEx96-S5 ocean acoustic experiments demonstrate the effectiveness of the proposed methods. The results indicate that the proposed algorithms can achieve CRLB performance at low noise levels, with the TWLS algorithm showing the lowest complexity, while the Adam algorithm achieves the highest accuracy. These proposed solution algorithms can be selected based on computational complexity, localization accuracy and source of interest location.

Quantum sensors often consist of packaging, such as dielectric-based vapor cells and metallic electrodes, that reduces and spatially alters the locally observed electromagnetic fields. These effects have been well studied in the optical regime, and even in the RF regime over a few GHz. However, there have been few studies in the electrically small regime below 1 GHz. In order to account for or remove the effects of the packaging, more studies are needed across a broad range of frequencies. This paper reports on the complex permittivity and conductivity of several commercially available vapor cells used for Rydberg electric field sensing from 10-300 MHz. A new method using a stripline transmission measurement was performed and full wave electromagnetics modeling was used to extract the effective dielectric constitutive parameters from the vapor cells. Additionally, the field reduction inside the vapor cell is reported, and published atomic measurements of the electric field are used to further validate the results presented here. Several observations were made from the measurements, such as the frequency dependencies of the RF dispersion and absorption. Applications of this technique include making precise numerical field corrections or physically designing a more optimal vapor cell via coatings, material changes, or geometric changes to improve field strength and uniformity.

The capacity for sustained contextual coherence in language model outputs remains a critical factor in achieving natural, relevant, and semantically rich responses across extended interactions. Traditional context-handling methods, often limited to token-level attention and basic memory retention mechanisms, fall short when confronted with the intricate, hierarchical nature of real-world discourse. Multi-Stage Latent Contextual Synthesis presents a novel solution, employing a structured, multi-tiered synthesis approach to refine context integration within an LLM framework. Through a sequential layering process, latent knowledge is synthesized at progressively deeper levels, yielding a more nuanced understanding of semantic dependencies, thematic continuity, and response relevance across multi-turn conversations. The experimental results reveal substantial improvements in contextual retention, thematic alignment, and token stability, underscoring the method's efficacy in enhancing complex context management without adding prohibitive computational demands. Comprehensive evaluations demonstrated that Multi-Stage Latent Contextual Synthesis not only enhances memory efficiency and reduces thematic error rates but also promotes embedding stability, thereby supporting a more structured and responsive approach to contextual synthesis in high-stakes applications, including conversational AI and automated support systems. These findings illustrate the potential of hierarchical, latentcontextual strategies to redefine contextual processing standards within LLMs.

The next generation of Wi-Fi is meant to achieve ultra-high reliability for wireless communication. Several approaches are available to this extent, some of which are being considered for inclusion in standards specifications, including coordination of access points to reduce interference. In this paper, we propose a centralized architecture based on digital twins, called WiTwin, with the aim of supporting wireless stations in selecting the optimal association according to a set of parameters. Unlike prior works, we assume that Wi-Fi 7 features like multi-link operation (MLO) are available. Moreover, one of the main goals of this architecture is to preserve communication quality in the presence of mobility, by helping stations to perform reassociation at the right time and in the best way.

Vehicle-to-vehicle (V2V) communications require highly accurate and real-time channel modeling techniques. The dynamic nature of V2V scenarios demands continuous scene updates while ray-tracing (RT) algorithms typically require substantial computational time to achieve accurate predictions. This letter proposes an OptiX-based GPU-parallel dynamic shooting and bouncing ray tracing (DSRT) method that combines the advantages of shooting and bouncing ray (SBR) and RT approaches. The method employs an enhanced adaptive ray tube splitting mechanism to improve accuracy. A level of detail (LOD)-based dynamic scene update method further optimizes performance, while OptiX and CUDA acceleration enable efficient RT. Channel measurements in urban environments demonstrate the high accuracy of our algorithm. Performance evaluation results show that DSRT achieves a parallelization degree exceeding 98.8% and delivers exceptional speedup ratios.

Power cycling reliability is one of the most rigorous methods to evaluate the durability of power semiconductor devices against real-life varied electrical and thermal stresses. The corresponding test results can generate lifetime models to predict the service life in operating conditions. This paper provides an overview of the power cycling reliability testing for power semiconductor devices, focusing on both legacy Si and new wide bandgap (WBG) devices. It begins by presenting the power semiconductor packaging trends, their evolving failure mechanisms, and the power cycling test methods. Different testing standards, lifetime models, and their challenges are then discussed. Special attention is given to WBG devices, examining various testing methods and their impacts. A generic lifetime model for different semiconductor materials and packaging technologies is provided based on above 200 samples collected from the existing literature, which can serve as a useful tool for reliability evaluation. Finally, the limitations and associated open questions are discussed to identify future research opportunities.

5G wireless networks harness the extensive spectrum available in the millimeter-wave (mmWave) frequency bands which set them apart from current wireless systems in terms of directivity, propagation loss, and susceptibility to blockages. Sub-6 GHz systems can attain omni-directional coverage, displaying limited sensitivity to physical obstacles. Still, they are incapable of achieving the same level of service quality as systems outfitted with electronically steerable directional antennas offering reduced propagation loss and higher gains due to the beam directionality. In our framework, we investigate the utilization of directional, steerable device-to-device (D2D) mmWave antennas as integral components. The nodes communicate by manipulating the orientation of their antennas, i.e., steering their beams. To minimize dependence on a base station, the D2D nodes are categorized into primary and secondary ones and they communicate in three phases: Uplink, Downlink, and PAPA. We delve into the impact of optimal steering of the main lobe beams transmitted by these D2D antennas as well as optimizing the time sharing among three phases. Within each phase, we assume that the nodes follow a random transmission scheduling scheme and derive the achievable rates accordingly. Through meticulous design of polynomial-time heuristics, we maximize the overall network capacity.

It is commonly known that gray matter, a tissue type with high concentrations of neuronal cell bodies, is indicative of brain health. Nonetheless, there has not been a lot of research on the relationship between gray matter volume and incidental memory. In my study, I analyzed the anatomical images of the participants of the Magic, Memory, and Curiosity research project, a study dedicated to the influence of magic tricks on a subject's curiosity and incidental memory. Using the provided masks, I utilized FSL to calculate the ratio of gray matter volume to total brain volume. After doing so, I calculated the p-value and r-value for the correlation between the volume ratio and the eight different memory scores. The recollection and recall scores turned out to be a moderate combination of positive and negative correlation coefficients, and an appreciable number turned out to be statistically significant. It follows that there is a correlation between gray matter volume and recall as well as recognition memory. In the future, scientists could measure the volume ratio of a brain region specifically associated with memory, such as the hippocampus or amygdala. Perhaps, measuring the gray matter thickness or using another set of measures would have also yielded more significant results.

Chronic low back pain (cLBP) profoundly impacts quality of life, yet its underlying mechanisms remain poorly understood. Three-dimensional (3D) ultrasound imaging offers valuable insights into cLBP but poses challenges of manual annotation due to large data volume and poor image quality. To address these issues, a Cross-Scan Fusion Network (CSFN) is proposed for anatomical tissue layer segmentation in 3D ultrasound images. CSFN leverages a VoxelMorph-based registration model with Projected Hausdorff Loss (PHDL) to generate synthetic image-mask pairs, enabling both sample-efficient learning (CSFN-SEL) and semi-supervised learning (CSFN-SSL). Our results demonstrated that CSFN-SEL achieved superior segmentation performance compared to fully supervised nnU-Net (baseline) across varying labeled samples, with mean Dice coefficients of 74.62% ± 2.45% (5 samples) and 80.93% ± 1.87% (19 samples). CSFN-SSL further improved performance, achieving 79.33% ± 1.96% and 81.14% ± 1.43% for 5 and 19 samples, respectively, significantly outperforming baseline methods (p < 0.05). These findings highlight the effectiveness of CSFN in enhancing segmentation accuracy and robustness under limited labeled data conditions, offering a feasible solution for advancing cLBP imaging and analysis.