The ever-growing digital repositories of medical data provide opportunities for advanced healthcare by forming a foundation for a digital healthcare ecosystem. Such an ecosystem facilitates digitized solutions to aspects like early diagnosis, evidence-based treatments, precision medicine, etc. Contentbased medical image retrieval (CBMIR) plays a pivotal role in delivering advanced diagnostic healthcare within such an ecosystem. The concept of deep neural hashing (DNH) is introduced with CBMIR systems to aid in faster and more relevant retrievals from such large repositories. The fusion of DNH with CBMIR is an interesting and blooming area whose potential, impact, and methods have not been summarized so far. This survey attempts to summarize this blooming area through an in-depth exploration of the methods of DNH for CBMIR. This survey portrays an end-to-end pipeline for DNH within a CBMIR system. As part of this, concepts like the design of the DNH network, utilizing diverse learning strategies, different loss functions, and evaluation metrics for retrieval performance are discussed in detail. The learning strategies for DNH are further explored by categorizing them based on the loss function into pointwise, pairwise, and triplet-wise. Centered around this categorization, various existing methods are discussed in-depth, mainly focusing on the key contributing aspects of each method. Finally, the future vision for this field is shared in detail by emphasizing three key aspects: current and immediate areas of research, realizing the current and near-future research into practical applications, and finally, some unexplored research topics for the future. In summary, this survey depicts the current state of research and the future vision of the field of CBMIR systems with DNH.

Commercial Unmanned Aircraft Systems (UAS) have a wide range of applications, including package delivery, inspection as well as search and rescue missions. To operate Unmanned Aircraft Vehicles (UAV) Beyond Visual Line of Sight (BVLOS), reliable long-distance communication is essential. Although the cellular network is a potential solution, it still faces issues such as signal loss and increased handover at higher altitudes. To mitigate these issues, our work proposes the usage of two cellular links from different providers, which are prioritised according to Quality of Service (QoS) prediction. We evaluate multiple AI-based model architectures for the prediction, and find that the model consisting of Gated Recurrent Units (GRU) and convolutional layers outperforms the others. The models are trained and tested on real-world data showing a reduction in latency peaks, thereby increasing connection resilience. Additionally, the prediction pipeline is designed to be executable on the UAV side and is not limited to a specific geographical area, making it applicable to real-world scenarios. Finally, we present a pre-flight path planning algorithm that considers QoS when calculating the flight path in order to further enhance the communication. To support the research community, we publicly share the dataset used to obtain our results.

The most pressing challenge in the practical application of chaotic systems is the development of methods for encrypting information. Addressing these challenges requires the creation of chaotic oscillation generators with high complexity indices. This paper introduces a novel 4-dimensional (4D) memristive system that is characterized by its simplicity, consisting of only seven terms and lacking equilibrium points, making it capable of generating hidden attractors. The dynamic properties of the system are thoroughly analyzed, including bifurcation diagrams, Lyapunov exponents, Kaplan-York dimensions, and offset boosting analysis. Furthermore, the theoretical model is validated through an electronic simulation of the new two-winged chaotic system using Multisim. Synchronization between two identical 4D hyperchaotic systems is successfully achieved using an active control method.

Due to electrification, magnetic sensors are increasingly deployed in magnetically polluted environments. To reject stray fields, differential sensing schemes are typically used. In this paper, we rely instead on a single-point gradiometric sensing scheme based on the force exerted on a magnet, which is directly related to the magnetic field gradient. Unlike the prior art, our concept uses a differential MEMS force-sensing mechanical transducer. The overall sensor rejects not only magnetic stray fields, but also mechanical disturbances such as vibration and gravity. It is the first single-point MEMS gradiometer that can operate unshielded in various orientations. Our prototype realization achieves a noise density of 4 nT/mm/√Hz in a ±300 μT/mm measurement range. We demonstrate its operational effectiveness in a bus-bar current sensing application. Finally, the paper discusses the remaining weaknesses and provides an outlook for MEMS-based magnetic sensors.

In order to achieve the ubiquitous connectivity expected by 6G wireless networks, in this paper, an artificial intelligence (AI) framework is proposed for integrating intelligent omni-surface (IOS) into the cell-free (CF) network while leveraging millimeter wave frequency bands. The proposed AI framework allows the wireless network to allocate the necessary power for beamforming from the required base stations (BSs) to serve the users positioned on both sides of the IOS, thereby extending the wireless system's coverage area. An optimization problem is formulated to maximize the achievable rates of the users, thereby maximizing the signal-to-interference-plus-noise ratio (SINR) while saving power to support the maximum number of users in the proposed CF network. To solve the formulated NP-hard problem, a novel AI-based framework is proposed that allocates power to users on both sides of the IOS, enabling fulldimensional coverage. In particular, a deep deterministic policy gradient (DDPG)-based deep reinforcement learning approach is employed to allocate the required power for beamforming from the necessary BSs in the CF network, effectively serving users on both sides of the IOS. Simulation results show that the proposed DDPG-based framework outperforms state-of-the-art baselines such as advantage actor-critic and deep Q-network methods, achieving cumulative SINR improvements ranging from 5.21 dB to 5.23 dB and 6.43 dB to 6.96 dB, and cumulative achievable rate enhancements ranging from 5.45 bps/Hz to 5.47 bps/Hz and 6.67 bps/Hz to 7.20 bps/Hz, correspondingly, considering users only on one side and on both sides of the IOS.

Without a low-cost, highly efficient power conversion circuit system, the current achievements in Information and Communication Technology would not have been possible. The integration of power conversion circuit systems began with data writing for non-volatile semiconductor memories. Utilizing a switched-capacitor converter has enabled full integration due to its configuration of switch, capacitor, oscillator, and regulator components. With the scaling of CMOS technology, the supply voltage for LSIs has started to decrease. The voltage drop across switches has relatively increased, resulting in escalating costs and power losses. To address this, a shift from passive switches to active ones has been made to mitigate voltage drops and ensure scalability. Reconfigurability has been developed to maintain low cost and high power-efficiency even in applications with significant fluctuations in voltage conversion ratios. On the other hand, digital circuits used to operate on power supplies downscaled inefficiently by linear regulators. As the integration density of digital circuits reached the power density limits, more efficient switched-capacitor voltage-down converters have become prevalent to enhance power efficiency. Nowadays, power supply voltages and clock frequencies are adjusted dynamically based on computational loads. Thus, reconfigurability remains crucial even in switched-capacitor voltage-down converters due to demands for variable voltage conversion ratios. In this article, we explore the evolution of reconfigurable switched-capacitor converters, tracing their development from the rich history of switched-capacitor converters spanning over a century.

In this paper, we propose an address-dependent divided-bit-line (BL) NAND Flash memory for reduction in read latency. Energy in bit-line path can be also reduced with this design. A single switching MOSFET is added per BL. When a memory cell located closer to the sense amplifier is selected for read, the switch gets turned off. The sense amplifier sees shorter BL with smaller parasitic capacitance and resistance, resulting in lower latency and energy in BL path. When a memory cell located farther from the sense amplifier is selected for read, the switch gets turned on. The sense amplifier sees the original bit-line when the on-resistance of the switch is sufficiently low in comparison with the BL resistance. The averaged BL access time and energy can be minimized analytically when the switch is placed at xSW of 0.58 and 0.50, respectively, where xSW indicates the switch location (xSW = 0 at the closest to the sense amplifier and xSW = 1 at the farthest from the sense amplifier), assuming that the memory cells are accessed in random. With the proposed design, the average read latency and energy can be reduced to 62 % and 75 %, respectively. The effectiveness of the proposed design was validated with SPICE. This paper also investigates 1) the case where hot memory cells are placed in the closer side whereas cold cells are in the farther side and 2) the case where more switches are built in every BL. In this paper, NAND Flash memory is considered to quantifying the effectiveness of the proposed design on read latency and energy.

Let S∞/Z2 be the infinite lens space. Denote the Steenrod algebra over the prime field F2 by A . The (mod-2) cohomology H∗ ((S∞/Z2) ⊕s ; F2) is known to be isomorphic to the graded polynomial ring Ps := F2[x1, . . . , xs] on s generators of degree 1, viewed as an unstable A -module. The Kameko squaring operation (Sqg0 ∗ )(s;N) : (F2 ⊗A Ps)2N+s −→ (F2 ⊗A Ps)N is rather useful in studying an open problem of determining the dimension of the indecomposables (F2 ⊗A Ps)N . It has been demonstrated that this (Sqg0 ∗ )(s;N) is onto. As a continuation of our recent works, this paper deals with the kernel of the Kameko (Sqg0 ∗ )(s;Nd) for the case where s = 5 and the "generic" degree Nd is of the form Nd = 5(2d − 1) + 11.2 d+1 for arbitrary d > 0. We then rectify almost all of the main results that were inaccurate in an earlier publication [Rev. Real Acad. Cienc. Exactas Fis. Nat. Ser. A-Mat. 116:81 (2022)] by Nguyen Khac Tin. We have also constructed several advanced algorithms in SAGEMATH to validate our results. These new algorithms make an important contribution to tackling the intricate task of explicitly determining both the dimension and the basis for the indecomposables F2 ⊗A Ps at positive degrees, a problem concerning algorithmic approaches that had not previously been addressed by any author. Also, the present study encompasses an investigation of the behavior of the cohomological transfer in bidegrees (5, 5+Nd), with the internal degree Nd mentioned above.

Twisted pair is a well-known and widespread media in modern wired communications. In this entry-level tutorial for both experts and many non-experts in the field, we are theorizing on why, what, and where such communications could gain thanks to an advance from the pair into a triplet of conductors.

This paper describes the properties of pin Ge1-ySny diodes (y=4.4-10% Sn) grown directly on Si(100) wafers as a way to investigate the impact of eliminating the Ge buffer layers used conventionally for the integration of GeSn devices on Si. The technology offers a simplified and potentially lower-cost alternative for SWIR-LWIR applications. Two device designs are discussed. The first design adopts a layer sequence n-Ge1-ySny/i-Ge1-ySny/p-Ge1-ySny/Si, featuring a single defected bottom interface between the p layer and the Si wafer. This was followed by an even simpler n-Ge1-ySny/i-Ge1-ySny /p-Si heterostructure design. In both cases, the top i/n interface is pseudomorphic and potentially defect-free. The Ge1-ySny layers are produced by CVD reactions of Ge3H8 and SnH4 at temperatures ranging from 300 o C to 390 o C. The n-type electrodes in the samples were doped with As using As(SiH3)3, and the p-type GeSn layers were doped using diborane as the source of B-atoms. All samples were characterized by XRD, RBS, IR-ellipsometry, AFM and TEM. The layers were found to be monocrystalline single-phase alloys exhibiting mostly relaxed strain states and top surfaces devoid of the cross-hatch surface patterns that are typical of Ge1-ySny films grown on Ge buffers. Current voltage I-V curves of fabricated devices over the 4.4-10% Sn range of interest showed that rectifying behavior is readily attained. It appears that the effect of eliminating the Ge-buffer is an increase of only one order magnitude in the density of defects responsible for the dark current, together with an increase in residual doping in the nominally intrinsic layer. The results suggest that these deleterious effects may be further reduced with improved sample designs, particularly at high Sn-concentrations, opening up new alternatives for the effective integration of GeSnand Si technologies.

As digital economies continue to grow and the demand for wireless communication intensify, efficient and effective spectrum management becomes increasingly vital. In response, regulatory bodies worldwide are interest in leveraging advanced technologies to enhance and expedite their spectrum management efforts. This paper investigates the application of large language models (LLMs) for regulatory processes and their potential to improve operational efficiencies and decision-making. In addition, this paper demonstrates how LLMs enhance and accelerate policy and regulation processes for information extraction and question answering; focusing on spectrum management use cases. The results obtained indicate a robust document processing alternative, suggesting a transformative impact on spectrum regulatory practices.

The development of autonomous vehicles (AVs) is rapidly transforming transportation, promising increased safety, efficiency, and accessibility. This paper presents an endto-end machine learning approach for autonomous vehicle control, encompassing perception, decision-making, and actuation. Unlike traditional modular approaches, which rely on separately designed components for tasks such as object detection, path planning, and control, our method integrates these stages into a unified model. Leveraging deep learning techniques, particularly convolutional neural networks (CNNs) and recurrent neural networks (RNNs), the model directly maps raw sensory inputs, such as camera images and LIDAR data, to driving commands. This approach allows the system to learn and adapt to complex driving environments, reducing the need for hand-engineered features and rule-based systems. The performance of the proposed model is evaluated through simulation and real-world testing, demonstrating its capability to handle diverse scenarios, including urban traffic, highway driving, and obstacle avoidance. The results highlight the advantages of an end-to-end strategy in terms of scalability, robustness, and generalization, marking a significant step toward fully autonomous vehicles.

Soccer is one of the most popular sports worldwide, with millions of fans and players. Teams often invest significant resources into improving their performance and achieving success in their respective leagues. One key aspect of a team’s performance is their playing style, which reflects their strengths and weaknesses on the field. Understanding how playing styles vary across different countries and leagues can provide valuable insights for coaches, managers, and analysts. Understanding how playing styles vary across different countries and leagues can provide valuable insights for coaches, managers, and analysts. In this project, we aim to analyze the playing styles of professional soccer teams from six different countries: Italy, Germany, Spain, England, France, and the United States. By using K-means clustering to group teams, we can understand how their playing styles differ from each other. We will use real datasets from the previous year and compare our results with past research papers to provide insights into the similarities and differences in playing styles across different countries and leagues. Our findings will inform coaching strategies and improve our understanding of the game.

This paper describes the auditory EEG challenge, organized as one of the Signal Processing Grand Challenges at ICASSP 2024. The challenge provides electroencephalogram (EEG) recordings of 105 subjects who listened to continuous speech, as audiobooks or podcasts, while their brain activity was recorded. The challenge consists of two tasks that relate EEG signals to the presented speech stimulus. The first task, called match-mismatch, is to determine which of five speech segments induced a given EEG segment. The second task, called regression, is to reconstruct the Mel spectrogram from the EEG. EEG recordings of 85 subjects were provided as a training set so that challenge participants could train their models on a relatively large dataset. The remaining 20 subjects were used as held-out subjects for the evaluation step of the challenge.

As Artificial intelligence (AI) transforms industry, the role of AI Product Managers (PMs) has evolved to navigate the complexities of AI-driven products. This research employs grounded theory to explore the evolving role of AI product managers (PMs), highlighting their unique responsibilities, challenges, skills, and key traits. AI PMs engage in extensive crossfunctional collaboration, data management, ethical AI development, and continuous learning. They face challenges such as ensuring data ethics, navigating AI regulations, managing the high cost of AI models, monitoring model performance, managing AI hype, and building user trust in AI products. Essential skills include technical proficiency, data literacy, critical thinking, and effective communication, and traits such as curiosity, growth orientation, and passion for AI are crucial. The study reveals how AI impacts the product lifecycle, such as by accelerating prototyping and extending the ideation and post-launch phases. This comprehensive analysis provides valuable insights into the dynamic role of AI PMs in a technology-driven landscape, and illuminates an archetype persona of an AI product manager.

Serverless enables users to execute the code and shift the management problem towards Cloud Service Providers. Resource management in serverless computing environments presents unique challenges and opportunities due to the fundamentally different operational model compared to traditional server-based architectures. Serverless computing abstracts the underlying infrastructure from developers, allowing them to focus on writing code rather than managing servers. This abstraction, however, introduces complexities in resource management, necessitating innovative approaches to optimize performance, cost, and scalability. In serverless environments, resources such as CPU, memory, and storage are dynamically allocated and deallocated by the cloud provider based on the workload. This dynamic nature can lead to unpredictable performance characteristics and cost implications, as the resource usage scales with the number of function invocations. Whenever a function is invoked it must be served by a virtual machine. The major objective behind the function execution is to schedule the incoming function to the most suitable VM so that the VM resources can be utilized efficiently. It offers real-time scheduling of function instances that need to be executed to obtain the best performance and significantly lowers energy consumption. The study provides an extensive overview of resource management strategies in a serverless setting within this framework. Regarding cloud computing, a conceptual framework for workload prediction and resource management, classification of current machine learning-based resource allocation techniques, and key issues with inefficient physical resource distribution are covered. After that, a thorough analysis of current methods supporting machine learning-based methods in serverless resource management is provided. Ultimately, the study examines and summarises several recently identified issues as well as potential future research areas related to elastic resource management in serverless computing environments.

Degradation of energy capacity and energy efficiency of lithium-ion battery energy storage systems depend on their operation conditions and are key factors in their business case feasibility. Using traditional energy reservoir models with energy throughput methods to quantify degradation leads to less accurate revenue estimates and unfeasible dispatch scenarios. This paper proposes a new neural network-based model of a lithium-ion battery energy storage system that replicates both degradation process and energy efficiency as functions of the operating conditions. The neural networks are trained using datasets derived from a detailed physics-based digital twin of lithium-ion cell models. The architecture of the neural networks is designed to allow their straightforward integration into mixedinteger optimization frameworks that are often used for energy storage systems simulations. The proposed model is tested for long-term energy arbitrage studies and the findings support the effectiveness of the proposed model, leading to extended lifespan and higher revenues for energy arbitrage studies when compared to conventional models.

Digital transformation is essential for contemporary businesses, promising enhanced efficiency, customer satisfaction, and competitive edge. However, conventional metrics like Return on Investment (ROI) often fall short in capturing the full spectrum of benefits and challenges inherent in digital transformation initiatives. This article critiques traditional ROI metrics and explores alternative measures for assessing the success of digital adoption and change management efforts. By integrating qualitative indicators such as employee empowerment, agility, and cultural alignment with quantitative metrics, this study aims to provide a comprehensive evaluation of digital transformation’s impact on organizational resilience and competitiveness.