This paper introduces Atelier, a large language model (LLM)-based framework for analog circuit design to address the issues of data scarcity and the substantial domainspecific knowledge required in this field. Atelier integrates general-purpose LLMs with a high-quality, compact knowledge base to fulfill the considerable knowledge requirements of analog circuit design, obviating the need for extensive domain-specific training or fine-tuning. The knowledge base is meticulously curated to be task-oriented and encapsulates critical information from pertinent literature within user-defined templates, leveraging the LLMs' capabilities in text comprehension and summarization. The framework comprises several LLM agents, structured in a graph-of-thoughts architecture, with each agent specialized in a distinct task in analog circuit design, including circuit analysis, topology selection, topology modification, parameter tuning, and design decision. This collaborative multi-agent system, enriched with access to the compact knowledge base and advanced mechanisms such as self-reflection, backtracking, and tool integration, automates the analog circuit design process. It significantly enhances design quality and efficiency while ensuring interpretability. Experimental results highlight Atelier's superiority over state-of-the-art black-box methods, generalpurpose LLMs, and LLM-based methods, demonstrating notable improvements in success rates, design quality, and runtime.

In this research, we will build an algorithm (PCO: Particle Collision optimization) to solve continuous optimization problems. We assume that the optimal solution is confined to a limited space, and that a group of particles is scattered in the solution space according to a Brownian motion. These particles may collide with the optimal solution and may collide with each other, or even with the boundaries of the search space. We will test this metaheuristic algorithm on a set of standards benchmarks, and we will compare its results with the results of a group of wellknown algorithms, especially those that rely on the Lévy movement. We will also test its application to an engineering problem.

Antenna arrays with discrete antenna elements have been conventionally used in multiple-input multiple-output (MIMO) communications. Improved spectral efficiency resulting from more antenna elements has motivated the idea of packing very large numbers of elements in an array, yielding continuous-like antennas. This paper conducts a performance analysis of continuous antennas, employing matched filtering. We derive a gamma approximation for the instantaneous received signalto-noise ratio (SNR) under a Rician channel with a spatiallycorrelated non-line-of-sight (NLoS) component. We also derive the mean and an upper bound for achievable rate. Numerical results demonstrate the accuracy of our approximations across a range of parameters.

Edge case or long-tail detection in computer vision presents a significant challenge due to the inherent data imbalance and scarcity of rare event samples. Traditional convolutional neural networks (CNNs) often struggle to generalize well to such infrequent instances. In this paper, we propose a novel Transformer-based framework that effectively addresses the long-tail distribution problem by leveraging attention mechanisms to focus on subtle and rare patterns in images. Our approach incorporates a specialized loss function and data augmentation techniques to further enhance performance on edge cases. We evaluate our method on several benchmark datasets and demonstrate that it outperforms stateof-the-art models in detecting rare events, highlighting the potential of Transformers in tackling the long-tail problem in computer vision.

Accurate weather and climate prediction is essential for early warning systems that improve response strategies to climate-related events. This study explores the use of association rule mining (ARM) techniques to analyze large-scale meteorological datasets. We focus on the weather patterns of Tallinn and Tartu, investigating variables such as wind speed, temperature, precipitation, and humidity, and their influence on weather intensity. A distributed ARM approach is employed using the Apollo framework, which utilizes serverless functions to enhance scalability and performance. Results show Apollo outperforms traditional systems like Apache Spark by approximately 15% in terms of processing speed, while extracting a greater number of meaningful rules. Time-series analysis was also applied to investigate temporal weather trends. Our findings highlight the potential of this approach for enhancing weather prediction systems and offer a foundation for future research in this area.

This paper proves from scratch that the class of problems P is different from the class of problems NP. We define a sequence of sublanguages of the language SAT, which we will call languages of type A, consisting of logical formulas whose length is conditioned by the number and shape of the intersections of their clauses. As shown in the final sections, there exist formulas of that language of type A for which determining whether they are satisfiable is equivalent to determining whether the result of an arithmetic sum of a non-polynomial number of summands is greater than or equal to a natural number associated with each formula. It is shown that in order to obtain each of these summands at least one Turing machine step is required, which implies that this language of type A cannot be accepted by any deterministic Turing machine whose number of steps is bounded by a fixed polynomial as a function of the length of the input. It follows that this language of type A is not a language P but it is still a language NP, which leads to the conclusion that P is different from NP.

Current approaches to prosthetic control are limited by their reliance on traditional methods, which lack real-time adaptability and intuitive responsiveness. These limitations are particularly pronounced in assistive technologies designed for individuals with diverse cognitive states and motor intentions. In this paper, we introduce a framework that leverages Reinforcement Learning (RL) with Deep Q-Networks (DQN) for classification tasks. Additionally, we present a preprocessing technique using the Common Spatial Pattern (CSP) for multiclass motor imagery (MI) classification in a One-Versus-The-Rest (OVR) manner. The subsequent 'csp space' transformation retains the temporal dimension of EEG signals, crucial for extracting discriminative features. The integration of DQN with a 1D-CNN-LSTM architecture optimizes the decision-making process in real-time, thereby enhancing the system's adaptability to the user's evolving needs and intentions. We elaborate on the data processing methods for two EEG motor imagery datasets. Our innovative model, RLEEGNet, incorporates a 1D-CNN-LSTM architecture as the Online Q-Network within the DQN, facilitating continuous adaptation and optimization of control strategies through feedback. This mechanism allows the system to learn optimal actions through trial and error, progressively improving its performance. RLEEGNet demonstrates high accuracy in classifying MI-EEG signals, achieving as high as 100% accuracy in MI tasks across both the GigaScience (3class) and BCI-IV-2a (4-class) datasets. These results highlight the potential of combining DQN with a 1D-CNN-LSTM architecture to significantly enhance the adaptability and responsiveness of BCI systems.

Kolmogorov-Arnold Networks (KANs) have emerged as a powerful framework for modeling complex functions with minimal architecture, which can be a novel solution for Artificial Intelligence (AI) modeling. However, implementing KANs in hardware presents significant challenges due to their reliance on nonlinear functions. In this paper, we introduce the Stochastic Computing-based Kolmogorov-Arnold Network (SCKAN) Accelerator, the first known accelerator for KANs that leverages stochastic computing to effectively address these challenges. We propose a Segmented Multi-Dimensional Multi-Driving Finite State Machine (SMM-FSM) to efficiently approximate a broad range of nonlinear functions, enabling flexible and accurate network tuning. Our accuracy analysis demonstrates the feasibility of implementing KANs using stochastic computing. The SCKAN architecture integrates SMM-FSMs with crossbar networks, efficiently managing interconnections within the KAN. Through evaluations, we show that SCKAN achieves area reductions of 1.33 to 5.17× and power savings of 1.37 to 2.65× compared to traditional methods such as Look-Up Tables (LUTs), Taylor series expansion, and the Coordinate Rotation Digital Computer (CORDIC) algorithm. It maintains high accuracy in function-fitting tasks and minimal accuracy loss in classification tasks. SCKAN thus offers an effective solution for deploying KANs in resource-constrained environments, marking a significant advancement in the hardware implementation of nonlinear models.

Gastrointestinal or gastric cancer (GC) classification is a serious field of medical research and healthcare technology, where innovative machine learning (ML) and deep learning (DL) models are employed to categorize and analyze many kinds of GCs like pancreatic, gastric, or colorectal cancer. These models influence features extracted from medical imaging, genetic, and clinical data to distinguish between benign and malignant tumours, define cancer stages, and guide treatment verdicts. By automating cancer detection and classification procedures, DL techniques help healthcare experts make quicker and more exact analyses, leading to superior patient results, modified treatment tactics, and enhanced complete organization of GC cases. This technique has great promise for transforming initial detection and involvement in the battle against dangerous diseases. With this inspiration, this study presents a new snake optimization algorithm with a DL-assisted GC classification (SOADL-GCC) approach. The SOADL-GCC approach aims to examine the gastrointestinal tract images for the detection and classification of GC. To achieve it, the SOADL-GCC model employs a bilateral filtering (BF) approach for the noise removal process and enhances image quality. Besides, the SOADL-GCC technique uses a capsule network (CapsNet) model for deriving the feature vectors from preprocessed images. Moreover, SOA can achieve the optimum assortment of hyperparameters associated with the CapsNet model. Finally, the classification process can be performed using the deep belief network (DBN) model. A sequence of simulations took place on the Kvasir dataset to evaluate improved detection results of SOADL-GCC technology. An extensive comparative study reported that the SOADL-GCC technique effectively performs well with other models with a maximum accuracy of 99.72%.

Trajectory planning for automated vehicles in traffic has been a challenging task and a hot topic in recent research. The need for flexibility, transparency, interpretability and predictability poses challenges in deploying data-driven approaches in this safety-critical application. This paper proposes DeepGame-TP, a game-theoretical trajectory planner that uses deep learning to model each agent’s cost function and adjust it based on observed behavior. In particular, a LSTM network predicts each agent’s desired speed, forming a penalizing term that reflects aggressiveness in the cost function. Experiments demonstrated significant advantages of this innovative framework, highlighting the adaptability of DeepGame-TP in intersection, overtaking, car following and merging scenarios. It effectively avoids dangerous situations that could arise from incorrect cost function estimates. The approach is suitable for real-time applications, solving the Generalized Nash Equilibrium Problem (GNEP) in scenarios with up to four vehicles in under 100 milliseconds on average.

Ransomware attacks continue to evolve, with increasingly complex cryptographic payloads designed to evade detection and disrupt systems. A novel approach has been developed to address the challenge of evaluating the quality of true random bit-streams embedded within ransomware payloads, offering a fully automated method that eliminates the need for human intervention. Through entropy analysis and advanced randomness test suites, the methodology assesses the unpredictability and cryptographic strength of ransomware bit-streams, providing an objective measure of randomness. The findings reveal significant variability in the quality of randomness across different ransomware families, with some exhibiting potential cryptographic vulnerabilities. By leveraging mathematical models and statistical tools, the approach ensures scalable and efficient analysis, particularly important in high-volume ransomware datasets where traditional methods prove inadequate. The results have direct implications for improving ransomware detection and prevention strategies, particularly through identifying payloads that deviate from expected randomness levels. The research demonstrates the importance of randomness in ransomware encryption, offering valuable insights for both cybersecurity experts and automated detection systems.

Microservices-centric in-network computations, coupled with Named Data Networking (NDN) and supported by vehicular edge computing (VEC), offer a promising platform to meet the requirements of vehicular applications. However, a significant obstacle hindering high efficiency is the interdependency of microservices (MS), which may delay the timely results delivery due to pending interest table (PIT) timer expiration resulting in unsolicited packet drops. Moreover, voluminous data transmission in resource-constrained environment may prevent the consumer from receiving timely results. Therefore, to avoid unsolicited computation losses and enable proximate computations, this article envisions interdependent microservices offloading and semantic aware results transmission (iMSoRT) for 6G vehicular edge networks. iMSoRT formulates an efficient strategy that allows traffic controller (Tc) to account for MS interdependencies and allocate the PIT timer based on computational and network resource requirements. Moreover, iMSoRT introduces a live forwarding information base (lFIB) that effectively filters out underutilized RSU-E servers and forwards computation requests to the optimal RSU-E, accelerating processing while minimizing communication costs. Furthermore, iMSoRT develops a semantic transformer (ST) that leverages the YOLO and convolutional neural networks (CNN) and places it between the caching and forwarding module of NDN. The ST extracts and forwards semantically meaningful information, thereby reducing network resource utilization and enhancing information exchange efficiency. Simulation results revealed that iMSoRT achieved an impressive compute-hit ratio of over 90%, restrict cloud offloading to 12%, reduced computation delays over 50%, and optimized bandwidth utilization over threefold compared with benchmark schemes.

Functional near-infrared spectroscopy (fNIRS) is beneficial for studying brain activity in naturalistic settings due to its tolerance for movement. However, residual motion artifacts still compromise fNIRS data quality and might lead to spurious results. Although some motion artifact correction algorithms have been proposed in the literature, their development and accurate evaluation have been challenged by the lack of ground truth information. This is because ground truth information is time- and labor-intensive to manually annotate. This work investigates the feasibility and reliability of a deep learning computer vision (CV) approach for automated detection and annotation of head movements from video recordings. Fifteen participants performed controlled head movements across three main rotational axes (head up/down, head left/right, bend left/right) at two speeds (fast and slow), and in different ways (half, complete, repeated movement). Sessions were video recorded and head movement information was obtained using a CV approach. A 1-dimensional UNet model (1D-UNet) that detects head movements from head orientation signals extracted via a pre-trained model (SynergyNet) was implemented. Movements were manually annotated as a ground truth for model evaluation. The model’s performance was evaluated using the Jaccard index. The model showed comparable performance between train and test sets (J train = 0.954; J test = 0.865). It moreover demonstrated good and consistent performance at annotating movement across movement axes and speeds. However, performance varied by movement type, with best results for repeated (J test = 0.941), followed by complete (J test = 0.872), and then half movements (J test = 0.826). This study suggests that the proposed CV approach provides accurate ground truth movement information. Future research can rely on this CV approach to evaluate and improve fNIRS motion artifact correction algorithms.

The dominance of traditional paper-based ECG records in clinical settings poses a barrier to integrating AI for cardiovascular disease detection. The shift towards digital ECG is imperative yet hindered by the lack of reliable digitization tools. Our methodology comprises three key phases: lead segmentation, signal extraction, and scale conversion. Initially, we employ a pre-trained YOLOv8 single-shot detector to capture snapshots of 12 ECG leads. Then, each snapshot undergoes Otsu's technique for RGB to binary conversion, Hough transform for grid detection, and Viterbi's algorithm for sampling and extracting signal values in pixels. Finally, we convert these pixel values to voltage values by estimating grid resolution within the snapshot. To address uneven lengths in the extracted lead signals, we use cubic spline interpolation to horizontally scale the signals to align with a predetermined target length while preserving their waveform.

A control method for regulating both DC- and AC-side voltages, based on a disturbance observer, is presented. This method provides a voltage-source behavior from the AC-side perspective, a key functionality of grid-forming converters. A dynamic model is derived to develop the control law using feedback linearization. The control method is able to maintain the DC- and AC-side voltages without any cascaded loops. The method is therefore named the dual-voltage forming method to differentiate it from the recent definitions of grid-forming converters. The use of a disturbance observer provides integral action and also inherent synchronization. A transparent current controller is implemented to protect the converter from over-currents. Small-signal stability of the proposed method is studied analytically and design guidelines are drawn from the analysis. Furthermore, the asymmetric behavior of the converter in different operating modes is analytically assessed. The performance of the method is tested in experimental conditions using a 12.5-kVA test setup. The control method exhibits robust performance in both strong and weak grids in terms of DC-bus voltage reference tracking as well as the capacity to survive large external power variations with power-flow direction changes.

This paper introduces Med-Bot, an AI-powered chatbot designed to provide users with accurate and reliable medical information. Utilizing advanced libraries and frameworks such as PyTorch, Chromadb, Langchain and Autogptq, Med-Bot is built to handle the complexities of natural language understanding in a healthcare context. The integration of llamaassisted data processing and AutoGPT-Q provides enhanced performance in processing and responding to queries based on PDFs of medical literature, ensuring that users receive precise and trustworthy information. This research details the methodologies employed in developing Med-Bot and evaluates its effectiveness in disseminating healthcare information.

In recent years, DNN-based inversion of Ground Penetrating Radar (GPR) data has gained significant attention, primarily using simulated data for training due to the lack of ground truth in real-world scenarios. However, models trained on simulation data often perform poorly on real-world data due to domain discrepancies. This study applies an Unsupervised Domain Adaptation (UDA) method to GPR inversion, introducing a domain classifier within the main inversion DNN. By training the main DNN adversarially against the classifier to learn domain-invariant features, the DNN's performance can be improved even without labeled data in the target dataset. Several simulation datasets with varying characteristics were generated to compare the performance of three domain classifiers. The experiments revealed that the proposed simple domain classifier structure, using 3×3 convolutional local alignment, outperformed both pointwise local alignment and global alignment classifiers. Furthermore, the DNN was trained using unlabeled data from real-world roads measured by on-vehicle GPR, and then applied to a bridge deck slab specimen with known design drawings. From both qualitative and quantitative perspectives, the proposed method demonstrates superior performance in permittivity and material estimation. The method enabled clear detection of rebars, and damage such as deterioration and cracks in the concrete slab. This research is expected provide new insights into the efficient subsurface damage detection of road infrastructures.

Recently, the research community has recently shown overwhelming interest in metaverse-enabled wireless devices, due to their compelling proactive learning and selfsustainability attributes. Proactive learning enables machine learning models to be trained before user requests, while selfsustainability allows a system to function with the least amount of assistance from network administrators/users. Because of these features, one can use metaverse to enable various applications (e.g., entertainment and collision avoidance) in intelligent transportation systems. However, the limitations of computing processing power (e.g., in autonomous cars) and communication resources make implementing metaverse-empowered vehicular networks challenging. Motivated by these facts, we present a new framework for cooperative sensing, communication, learning, and task offloading for vehicular networks enabled by the metaverse. Subsequently, we formulate a cost-function minimization problem that accounts for transmission energy and transmission latency. The cost is minimized by optimizing task offloading, wireless resource distribution, transmit power allocation, and sensing interval. We employ a decomposition-based strategy for simultaneous resource allocation, task offloading, sensing interval optimization, and transmit power allocation. Due to the combinatorial nature of the resource allocation and task offloading problems, matching-based solutions are used. For sensing interval optimization, convex optimization is used. On the other hand, due to the non-convex and continuous nature of the transmit power allocation problem, a proximal term is introduced into the objective function to approximate it as convex objective function, which is then solved using a convex optimizer. To gain further insights, the proposed scheme is supported by extensive numerical results.