The Riemann zeta function is the most important special function in the large family of zeta functions. The Euler product is an infinite product of many functions in the multiplicative number theory. A completely multiplicative function gives the Euler product representation of the Riemann zeta function. In this article, the author provides a simple proof of the Euler product for the Riemann zeta function.

We describe the design, construction, and characterization of paper-based devices to perform spirometry, a standard test for lung function assessment. In this device, the instantaneous flowrate of the incoming breath from a person gets transformed to a specific acoustic frequency. In this manner, the time-course of the person's breath profile is mapped to a timevarying acoustic signal. The captured acoustic signal can be converted to standard spirometry curves and relevant parameters can be extracted as we describe in this paper. We compared our device with commercially available devices and showed that these paper-based devices provide similar performance. These devices have the advantage of low-cost, and simplicity of operation compared to currently available commercial products.

Accurate face detection and subsequent localization of facial landmarks are mandatory in many computer vision applications, such as emotion recognition, age estimation, and gender identification. Thanks to advancements in deep learning, numerous applications were developed for human faces. However, many applications have to employ multiple models to accomplish various tasks. As a result, they require more memory usage and increased inference time. Also, less attention is paid to other domains, such as animals and cartoon characters. To address these challenges, we propose an input-agnostic face model, AnyFace++, to perform multiple face-related tasks concurrently. The tasks are face detection and prediction of facial landmarks for human, animal, and cartoon faces, as well as age estimation, gender classification, and emotion recognition for human faces. We trained the model using deep multi-task multi-domain learning with a heterogeneous cost function. The experimental results demonstrate that AnyFace++ generates outcomes comparable to those of cutting-edge models specifically designed for specific domains. We have made the source code and pre-trained model publicly available at https://github.com/IS2AI/AnyFacePP to bolster research in this area.

We propose a self-supervised learning and a foundation model for analog layout automation. The proposed training method resolves the problem of lack of annotated layout data, and provides the foundation model with the implicit general knowledge on various appropriate layout designs in detail for various downstream tasks. For example, from the context of an incomplete layout pattern, a model fine-tuned from the foundation model can recommend the appropriate next design step toward the final layout. In experiments, vias, contacts, dummy fingers, and N-wells were successfully generated by models finetuned from the same foundation model, showing its tremendous potential.

During my time as a theoretical researcher at the quantum laboratories of the University of Paris-Sud in Orsay, France, in the late 1980s, I had the opportunity to meet one of the most prominent writers on simplifying quantum cybernetics and machine language, Mr. Viktor Davidovich Pekelis, a writer, specializing in quantum cybernetics who was visiting at the time for the Orsay Research Center (University of Paris-Sud).It was at this exact stage that the idea for this article began, but it did not take its current final form until 2023 / passing through a draft that was published in the TRONIX Magazine in Arabic in 2005.

This study explores the application of the YOLOv8 segmentation model in automated dental diagnostics through deep learning techniques, focusing on its effectiveness in detecting and classifying various dental features from X-ray images. The research emphasizes methodological rigor in dataset configuration, training with a pre-trained model, and employing advanced data augmentation strategies. Visualizations including bar charts and density plots illustrate the distribution and spatial characteristics of dental instances, highlighting prevalent categories like fillings and caries. The integration of deep learning in dental imaging not only enhances diagnostic precision but also supports personalized medicine by adapting diagnostic tools to individual patient data. This advancement holds promise for improving efficiency in disease detection and treatment planning, potentially transforming dental care delivery through enhanced automation and accuracy.

Software engineers integrate third-party components into their applications. The resulting software supply chain is vulnerable. To reduce the attack surface, we can verify the origin of components (provenance) before adding them. Cryptographic signatures enable this. This article describes traditional signing, its challenges, and changes introduced by next generation signing platforms.

This review systematically examines the progression of the You Only Look Once (YOLO) object detection algorithms from YOLOv1 to the recently unveiled YOLOv10. Employing a reverse chronological analysis, this study examines the advancements introduced by YOLO algorithms, beginning with YOLOv10 and progressing through YOLOv9, YOLOv8, and subsequent versions to explore each version's contributions to enhancing speed, accuracy, and computational efficiency in real-time object detection. The study highlights the transformative impact of YOLO across five critical application areas: automotive safety, healthcare, industrial manufacturing, surveillance, and agriculture. By detailing the incremental technological advancements in subsequent YOLO versions, this review chronicles the evolution of YOLO, and discusses the challenges and limitations in each earlier versions. The evolution signifies a path towards integrating YOLO with multimodal, context-aware, and General Artificial Intelligence (AGI) systems for the next YOLO decade, promising significant implications for future developments in AI-driven applications.

Software security is critical, yet challenges persist despite available tools and processes. This article explores secure pre-coding frameworks and strategies in the Software development lifecycle. Security first mindset necessitates collaboration among engineers, companies, and clients from a socio-technical perspective to ensure comprehensive secure software development.

Network digital twins (NDT) are transformative technologies that replicate physical network infrastructures in a virtual environment. These virtual models integrate real-time data feeds and advanced predictive analytics to simulate and forecast network behaviors, thus facilitating proactive management and decision-making processes. This technology underpins critical improvements in operational efficiency, reliability, and innovation in network management. By offering insights into potential failures and system efficiencies before they manifest in the physical world, NDTs enable stakeholders to make more informed, data-driven decisions. This paper explores the intricate framework of NDTs, their application across diverse industries, and the challenges they face, emphasizing their role in optimizing network performance and driving the next wave of digital transformation.

This paper investigates the transformative role of 6G networks in revolutionizing short video streaming, a sector that has witnessed exponential growth due to platforms such as TikTok, Instagram Reels, and Snapchat. With 6G's potential to deliver data speeds up to 100 times faster than 5G, coupled with near-zero latency and advanced network capabilities like edge computing and network slicing, the streaming of short videos is expected to achieve unprecedented quality and efficiency. We explore how these technological advancements will facilitate ultra-high-definition (UHD) streaming, enhance real-time interactivity, and enable deeper personalization through AI-driven analytics. Additionally, the paper discusses the implications for content creators and consumers, including new opportunities for engagement and the emergence of innovative business models, as well as the broader impacts on global media distribution networks. The integration of these technologies promises not only to refine user experiences but also to challenge existing media distribution paradigms, calling for novel regulatory and security frameworks to navigate the enhanced connectivity and data management capabilities of 6G networks.

The fifth generation (5G) communication, which is capable of realizing wireless communication with a extremely low latency and high data rate, has been put into commercial use widely. The propagation channel is of great significance for design, performance evaluation, and improvement of 5G system. As to investigate the wireless channel of 5G network, a channel measurement system based on passive sounding method is proposed in this paper. By utilizing the 5G downlink signal, e.g., synchronization signal block (SSB) and channel state information reference signal (CSI-RS), the proposed system is able to extract the multipath parameters of 5G propagation environment. Moreover, a measurement campaign is conducted at an urban macro-cellular (uMa) environment with the passive sounding system. The measured results is briefly analysed and compared with the 3rd generation partnership project (3GPP) results.

Bluetooth Low Energy (BLE) has emerged as a strong candidate for numerous Internet of Things (IoT) applications. Its advertising mode simplifies the complexity of implementation and extends the lifetime of equipment by eliminating the need for connection function, making it widely applicable in remote monitoring scenarios. In this research, we investigate the age of information (AoI) for monitoring over a BLE advertising network in which the duty cycle of the receiver scanning channels is set to 1 and each monitoring sensor broadcasts its status updates through m advertising events. Specifically, both the single-user and multiuser scenarios are examined. For the single-user scenario, we derive the average AoI and energy efficiency (EE) for the time-triggered (TT) traffic and eventtriggered (ET) traffic, respectively. On this basis, a comparative analysis is conducted between the average AoI of TT traffic and ET traffic, revealing that the average AoI of TT traffic is smaller than that of ET traffic with the same input rate. Furthermore, we explore the tradeoff between AoI and EE through minimizing the average AoI-EE ratio and minimizing AoI under EE constraints, respectively, by optimizing m. For the multiuser scenario, we characterize the receiving state through collision model, and investigate the AoI performance under zerowaiting policy. Further, we formulate a convex optimization problem to optimize the average AoI by adjusting the advertising interval. We not only propose an algorithm based on bisection method to find the exact solution, but also derive an approximate closed form solution.

Hyperspectral imaging (HSI) has proved to be quite potent when it comes to the analysis of documents which contain handwriting made from various types of ink and pen. Various computational techniques used on extracted iVision HHID mapped HSI cube to derive useful information is discussed in this paper. Targets outlined here are: number of spectral bands, representation of some bands as grayscale images, spectral responses of text pixels, registering of differing inks, classifying ink colours, and eventually, the extraction of features for enhanced dimensionality.

Gradient boosting is a widely used machine learning algorithm for tabular regression, classification and ranking. Although, most of the open source implementations of gradient boosting such as XGBoost, LightGBM and others have used decision trees as the sole base estimator for gradient boosting. This paper, for the first time, takes an alternative path of not just relying on a static base estimator (usually decision tree), and rather trains a list of models in parallel on the residual errors of the previous layer and then selects the model with the least validation error as the base estimator for a particular layer. This paper has achieved state-of-the-art results when compared to other gradient boosting implementations on 50+ tabular regression and classification datasets. Furthermore, ablation studies show that MSBoost is particularly effective for small and noisy datasets. Thereby, it has a significant social impact especially in tabular machine learning problems in the domains where it is not feasible to obtain large high quality datasets.

Steganalysis, the process of finding secret information within JPEG images. It plays a vital role in digital forensics and security. JPEG steganalysis uses various techniques to identify concealed information, mostly using statistical methods. However, due to advanced steganographic algorithm the JPEG steganalysis techniques will not be able to identify the secret information within JPEG images. Therefore, Adaptive JPEG steganalysis techniques are used. Adaptive JPEG steganalysis techniques and transfer learning provide tailored detection procedures, however, these approaches encounter difficulties in achieving model generalization and stability across various operational situations. In this paper, we propose Siamese Augmented Network (SAuGNet) an innovative approach for improving the efficiency of JPEG steganalysis by leveraging the fundamental concept of Siamese CNN-based architecture in combination with self-augmented convolution. This design improves the extraction and analysis of global features such as statistical features and transform coefficients across different parts of the image using self-augmented convolution. Experimental results show that the model achieves a detection performance that is competitive with existing steganalysis models. We conducted experiments on the BOSSBase and BOWS2 datasets, selecting samples randomly from each. These experiments were conducted across distinct scenarios, incorporating JPEG quality factors such as 75, 85, and 95, and embedding rates (bpnzAC) ranges from 0.1 to 0.5. The effectiveness of SAuGNet demonstrates superior detection capabilities compared to current steganalysis methods.

The recognition of voice commands in high-noise driving environments presents unique challenges that require implementing advanced strategies to ensure reliable and accurate performance. This paper introduces a method to improve the accuracy of voice command recognition in those environments. This study addresses the need for advanced strategies in the development of robust speech recognition systems for underresearched languages, such as Brazilian Portuguese, by optimizing speech-to-text results and introducing a new tool, the Audio Dataset Creator. It simplifies the creation of audio datasets and automatically validates all recorded audio through autogenerated transcripts. The study also emphasizes the potential impact of voice command recognition for driver assistance, particularly in enhancing accessibility and safety in vehicles adapted for drivers with reduced mobility. A novel Brazilian Portuguese audio dataset for vehicle dashboard control was created to develop the voice assistant system. This method demonstrated superior performance when compared to other speech recognition tools, correctly recognizing up to 99% of the voice commands, with only a small set of data for training. Furthermore, our transcriptbased command estimation approach shows an average 30% improvement in recognition accuracy. These results demonstrate the efficacy of the proposed method and tools in advancing voice command recognition technology for onboard automotive applications, which require lightweight and efficient solutions.

The tools and practices for software development have made substantial improvements over the course of the past decades. The subsequent generations of languages, frameworks and associated coding practices have evolved to improve readability, maintainability and to make the code less errorprone. One of the increasingly prominent paradigms in the latest generation of software development ecosystems is the declarative coding style. From Single Page Applications (SPA) with React, through Infrastructure as Code using Terraform, to declarative idioms in Scala and Haskell, this technique has been increasingly successful in enabling fast development of industrial quality software at scale. In this paper, we introduce PyDeclarative a-a new SPA framework for Python which inherits the best elements of Shiny/Shiny for Python and Qt Modeling Language (QML) in order to allow inline mixing of UI and logic elements; seamless reactive programming with minimal added verbosity; and compositional syntactic uniformity which delivers unparalleled flexibility in Dependency Injection/Inversion of Control mechanisms and refactoring capabilities. We compare our framework against state of the art alternatives in terms of features and performance.