Load forecasting plays a crucial role in energy management, directly impacting grid stability, operational efficiency, cost reduction, and environmental sustainability. Traditional Vanilla Recurrent Neural Networks (RNNs) face issues such as vanishing and exploding gradients, whereas sophisticated RNNs such as Long Short-Term Memory Networks (LSTMs) have shown considerable success in this domain. However, these models often struggle to accurately capture complex and sudden variations in energy consumption, and their applicability is typically limited to specific consumer types, such as offices or schools. To address these challenges, this paper proposes the Kolmogorov-Arnold Recurrent Network (KARN), a novel load forecasting approach that combines the flexibility of Kolmogorov-Arnold Networks with RNN's temporal modeling capabilities. KARN utilizes learnable temporal spline functions and edge-based activations to better model non-linear relationships in load data, making it adaptable across a diverse range of consumer types. The proposed KARN model was rigorously evaluated on a variety of real-world datasets, including student residences, detached homes, a home with electric vehicle charging, a townhouse, and industrial buildings. Across all these consumer categories, KARN consistently outperformed traditional Vanilla RNNs, while it surpassed LSTM and Gated Recurrent Units (GRUs) in six buildings. The results demonstrate KARN's superior accuracy and applicability, making it a promising tool for enhancing load forecasting in diverse energy management scenarios.

Explainable Artificial Intelligence (XAI) emerged to reveal the internal mechanism of machine learning models and how the features affect the prediction outcome. Collinearity is one of the big issues that XAI methods face when identifying the most informative features in the model. Current XAI approaches assume the features in the models are independent and calculate the effect of each feature toward model prediction independently from the rest of the features. However, such assumption is not realistic in real life applications. We propose an Additive Effects of Collinearity (AEC) as a novel XAI method that aim to considers the collinearity issue when it models the effect of each feature in the model on the outcome. AEC is based on the idea of dividing multivariate models into several univariate models in order to examine their impact on each other and consequently on the outcome. The proposed method is implemented using simulated and real data to validate its efficiency comparing with the a state of arts XAI method. The results indicate that AEC is more robust and stable against the impact of collinearity when it explains AI models compared with the state of arts XAI method.

This paper presents a novel framework for enhancing transfer learning capabilities in artificial intelligence agents, focusing on the crucial challenge of knowledge generalization across diverse tasks. We introduce the Adaptive Knowledge Transfer Network (AKTN), a hierarchical architecture that enables AI agents to decompose learned behaviors into fundamental cognitive primitives and recombine them for novel task execution. Our research demonstrates significant improvements in cross-domain knowledge application, reducing the learning curve for new tasks by up to 73% compared to traditional approaches.

The morphology and distribution of cell nuclei are critical for diagnosing diseases and predicting patient outcomes. However, the manual evaluation of histological images is often limited by low throughput and variability among observers. Deep learning models, such as U-Net, have been widely employed to address these challenges for semantic segmentation tasks in computational pathology. This study investigates the impact of varying encoder architectures in U-Net on the semantic segmentation of nuclear instances using the PanNuke dataset, a benchmark dataset for multi-class nuclear segmentation and classification. We evaluated six encoder architectures, including EfficientNet-B1 and Mix-ViT, and explored their performance in classifying and segmenting nuclear categories, such as tumor and non-tumor cells. The results demonstrated that EfficientNet-B1 achieved superior accuracy for classes with larger nuclei, such as epithelial cells. In comparison, Mix-ViT excelled in identifying smaller and scattered nuclei, such as inflammatory and connective tissue cells. Hyperparameter tuning and class-balanced data partitioning further enhanced the models' performance, mitigating the effects of class imbalance in the dataset.

The clothing industry demands unending novelty in attire designs to meet developing client advantages. However, established forms of founding attire patterns are labor-exhaustive, high-priced, and frequently impassable to tiny designers and arising fashion brands. FashionFusion presents an AI-compelled answer that automates the production of singular and superior attire patterns, leveraging the capacity of deep education and fruitful models. By streamlining the design process, this method allows designers to speedily original creative patterns, lowering two together opportunity and cost. The manifesto democratizes approach to progressive design finishes, enabling consumers to forge embodied designs outside the need for thorough mechanics knowledge. With this resolution, the impediments to fashion artistry are considerably reduced, promoting a more all-embracing and vital design environment.

In this project, we explore how to determine the safety of roads and intersections based on various factors like road connections, weather conditions, road quality, and traffic levels. While previous studies have used complex deep-learning techniques to predict traffic accidents, these methods can be hard for everyday Californians to understand and use. To address this, we propose a user-friendly web interface that helps people check the safety of the roads and intersections they travel most often. We trained a graph neural network using data on car accidents, road connections, weather, road conditions, and traffic volumes to assess safety more accurately, even when existing crash data is limited. This allows users to predict the safety of roads and intersections without having to wait for accidents to accumulate. The proposed web interface will be a Python-based web interface, created with the Taipy framework, which allows users to easily access and use the safety model.

This paper conducts a novel study of the P-Q coupling characteristics of grid-forming converters (GFM) under current-limiting and internal voltage-limiting in large disturbance scenarios. Initially, a simplified large-signal model is developed using internal voltage magnitude E and phase angle θ as key state variables. This model elucidates the nonlinear coupling between P , Q, E and θ, and examines how current and voltage thresholds affect the existence and accessibility of stable equilibrium points. By analyzing the trajectory evolution of operating points in the (θ, E) plane, the study uncovers the mechanisms and factors causing transient loss of synchronism in GFMs during phase angle jumps. It further details how internal voltage and currentlimiting influence the dynamic critical phase angle for loss of synchronization under the coupling effects of the P-f and Q-E control loops. Additionally, the paper addresses the dual effects of small-signal based P-Q decoupling control under large disturbances: enhancing resilience to minor phase jumps while potentially extending transient recovery. Hardware-in-the-loop tests confirm the theoretical analysis.

We show the Optical-Butterfly-Phenomena: the small differences in the structures of slits lead to profoundly different patterns.

In this work, we present single-event upset (SEU) analysis for Forksheet FET (FSFET) based CMOS circuits. Next, we present an array-level power and performance analysis along with the V min evaluation for the FSFET-based SRAM. Physicsbased TCAD and industry-standard BSIM-CMG compact models are calibrated for accurate circuit analysis in SPICE. The impact of varying Heavy-Ion Radiation (HIR) doses and strike orientations is investigated for the FSFETs. The robustness of CMOS inverter against HIR is also reported in terms of failure time (t fail) and output voltage swing (∆VDrop). For the SRAM, we determine the critical Linear Energy Transfer (LET). For FSFET, the individual n-/p-FETs are more vulnerable to the irradiation incident on nearby devices. At the circuit level, in comparison to perpendicular strikes, the ∆VDrop increases by 1.25 V and 2.75 V respectively, for oblique and transverse incidences, at a dose of 2.0 MeVcm 2 /mg. The t fail also increases by 43% and 60% and the SRAM critical LET also decreases by 85% and 57.5%, respectively. The array level SRAM evaluation shows that the FSFET enables reliable operation with low-power consumption, impressive noise margins, and low minimumoperating voltage (V min) values. FSFET SRAM power dissipation during the read and write operations is as low as 7.02 µW, and 3.00 µW respectively. At VDD=0.70 V, the noise margins for hold, read, and write operations are 289.27 mV, 122.89 mV, and 297.79 mV. The V min for read and write operations are 0.30 V and 0.35 V respectively.

Ground effect flight is able to seamlessly operate across highway, railway, water, and air corridors; but barriers emerge to on transfer to low-pressure tunnels and maglev. The barriers may be overcome with modes of operation that allow open entry to low pressure tunnels and higher speeds and efficiency with ground effect over rails than possible with maglev. Digital prototypes simulated in 2-Dimensional and 3-Dimensional computational fluid dynamics (CFD) have identified that open-entry low pressure tunnels can operate with seamless entry and exit with existing tunnel infrastructure and at speeds faster than airliners when operating with a favorable tunnel tailwind. Bernoulli Loops using tunnel air flow to transfer air between entry and exit locations are quantified with CFD simulations and create both resilience and degrees of multimodal optimization not possible with isolated tunnels. High-impact aspects of the multimodality include the ground-effect flight transit (GEFT) technology of this work including: a) seamlessness across modes from ranging from inner-city subways to open water, b) unprecedented increases in efficiency due to low drag and substantial elimination of rolling losses, c) higher speeds and higher efficiency than alternatives, d) a deployment strategy where inexpensive vehicles exhibit exceptional performance using existing infrastructure, and e) the rapid deployment of electric railcars using current infrastructure.

Current probes have long been used for Electromagnetic Compatibility (EMC) pre-compliance and diagnostic purposes. Dipole and transmission line-based models are often used to relate Common-Mode (CM) current levels to potential Radiated Emission (RE) levels. This relationship enables RE estimates to be performed without the use of accepted certification methodologies – reducing project costs and timelines. These models are based on a set of assumptions which cannot fully match real-world scenarios. This article expands the discussion by examining the relationship between the CM current and RE of a representative metallic structure using both measurements and a Finite Element Method (FEM) code. The relationship is shown to be frequency dependent, resulting in both over and under prediction of the RE levels. A practical recommendation is made on how to exploit the equivalence of CM current-based estimations of system RE.

Constrained environments, such as indoor and urban settings, present a significant challenge for accurate moving object positioning due to the diminished line-of-sight (LoS) communication with the wireless anchor used for positioning. The 5th generation new radio (5G NR) millimeter wave (mmWave) spectrum promises high multipath resolvability in the time and angle domains, enabling the utilization of multipath signals for such problems rather than mitigating their effects. This paper investigates the benefits of integrating multipath signals into 5G LoS-based positioning systems with onboard motion sensors (OBMS). We provide a comprehensive analysis of the positioning system’s performance in various conditions of erroneous 5G measurements and outage scenarios, which offers insights into the system’s behavior in challenging environments. To validate our approach, we conducted a road test in downtown Toronto, utilizing actual OBMS measurements gathered from sensors installed in the test vehicle. The results indicate that utilization of multipath signals for wireless positioning operating in multipath-rich environments (e.g. urban and indoor) can bridge 5G LoS signal outages, thus enhancing the reliability and accuracy of the positioning solution. The redundant measurements obtained from the multipath signals can enhance the system’s robustness, particularly when low-cost 5G receivers with a limited angle or range measurements are present. This holds true even when only considering the utilization of single-bounce reflections (SBRs).

In recent years, deep learning [1] has revolutionized various fields, such as computer vision, natural language processing [2], and speech recognition, by achieving state-of-the-art performance on numerous complex tasks [3]. These achievements are often attributed to the availability of large-scale, labeled datasets and the capacity of deep neural networks to learn rich feature representations. However, one critical challenge remains: the performance of deep learning models significantly deteriorates when there is a domain shift—a mismatch between the data distributions of the training set (source domain) and the test set (target domain). 

Decision trees are popular and highly interpretable machine learning algorithms used for both classification and regression tasks. These trees are structured with internal nodes representing features and leaf nodes indicating class labels or predictions. They are constructed through recursive partitioning based on the most informative features, employing criteria such as the Gini Index or Information Gain. Decision trees are appreciated for their interpretability, ease of use, and ability to handle both categorical and numerical features. They are robust against outliers and can effectively capture non-linear relationships. In this study, we explore the capabilities of decision trees by applying them to the MAGIC Gamma Telescope dataset to classify particles as gamma or hadron. By constructing a decision tree model, we aim to uncover the key factors and predictors for particle classification. The process includes data preprocessing, model construction, and evaluation to extract valuable insights and validate the model's predictive accuracy.

To maximize the autonomy of individuals with upper limb amputations in daily activities, leveraging forearm muscle information to infer movement intent is a promising research direction. While current prosthetic hand technologies can utilize forearm muscle data to achieve basic movements such as grasping, accurately estimating finger joint angles remains a significant challenge. Therefore, we propose a Multi-Stage Cascade Convolutional Neural Network with Long Short-Term Memory Network, where an upsampling module is introduced before the downsampling module to enhance model generalization. Additionally, we designed a transfer learning framework based on parameter freezing, where the pre-trained downsampling module is fixed, and only the upsampling module is updated with a small amount of out-of-distribution data to achieve transfer learning. Furthermore, we compared the performance of unimodal and multimodal models, collecting surface electromyography (sEMG) signals, brightness mode ultrasound images (B-mode US images), and motion capture data simultaneously. The results show that, on the validation set, the US image had the lowest error, while on the prediction set, the four-channel sEMG achieved the lowest error. The performance of the multimodal model in both datasets was intermediate between the unimodal models. On the prediction set, the average normalized root mean square error values for the four-channel sEMG, US images, and sensor fusion models across three subjects were 0.170, 0.203, and 0.186, respectively. By utilizing advanced sensor fusion techniques and transfer learning, our approach can reduce the need for extensive data collection and training for new users, making prosthetic control more accessible and adaptable to individual needs.

Reactive balance responses, which involve corrective and protective strategies, are highly dependent on rapid muscle activation to restore postural stability. Although electromyography (EMG) is commonly used to measure muscle activity, it has limitations such as signal interference, particularly during fast responses to external disturbances. Ultrasound imaging (US), in contrast, provides visualization of both superficial and deep muscles. Combining EMG and US imaging offers complementary insight into muscle behavior during reactive balance tasks. In this study, we investigated muscle activation and fascicle length changes in the medial gastrocnemius (MGS), lateral gastrocnemius (LGS), and soleus (SOL) muscles of the dominant (stepping) leg during stepping responses to unexpected low-amplitude (57.6 ± 5.8 N) and high-amplitude (123.4 ± 11.1 N) waist-pull perturbations in the anterior and posterior directions. Five young male adults (age: 25.2 ± 5.5 years) participated in the study. Results showed that perturbation amplitude significantly affected the EMG activation of both the MGS and SOL muscles in both directions, consistent with previous studies. Similarly, perturbation amplitude impacted fascicle length shortening in the LGS and SOL muscles. Significant differences in MGS and SOL activation were observed between high-amplitude and lowamplitude perturbations in both directions. Fascicle shortening in the LGS also differed significantly between perturbation amplitudes, whereas SOL fascicle shortening did not. By combining EMG and US imaging within the same participants, this study provides new insights into the neuromuscular mechanisms underlying balance control. These findings may inform the development of improved control strategies for neurorehabilitation devices and fall-prevention systems.

Differential semantic communication represents an intelligent model for meaning-based interactions between a transmitter and a receiver through a closed-loop feedback system. However, classical information theory, underpinned by Shannon's research, lacks the generality to analyse meaning-driven communications in complex feedbacked systems. To address this limitation, we propose a novel information theory for modern communication networks. Our theory introduces the concepts of transmitter and receiver capacities, which complement Shannon's channel capacity. We demonstrate how information redundancy can be strategically identified to align with communication goals. To validate the theory, we quantify key metrics to compare various image coding schemes. Our proposed work provides a theoretical foundation for advancing the design and optimisation of next-generation semantic communication systems.

We propose in this paper a novel hierarchical multi-grid orthogonal matching pursuit based angle estimation scheme. The proposed scheme accurately estimates the angles of departure (AoDs) and the angles of arrival (AoAs) of a point-to-point ultra massive multiple input multiple output (UM-MIMO) system, ensuring ultra-resolution estimation accuracy. Specifically, we define the on-grid error range in the conventional compressive sensing (CS) estimators and build high-resolution grids centered at the estimated discrete on-grid angles. The proposed scheme proceeds hierarchically through multiple levels to refine the angle estimations. Numerical simulations demonstrate the efficacy and superiority of the proposed scheme over recent state-of-the-art literature, where sub-millidegree angle estimation accuracy can be attained.