We discuss machine learning and its application to security, microchips and present a testable mathematical hypothesis on how the brain stores memories. In particular we present prototypes for next generation processors and RAM memories.

This survey provides an overview of the state-of-the-art in runtime adaptive Approximate Computing (AxC) for Deep Learning (DL) algorithms, highlighting the challenges and opportunities in the field. The survey covers a broad spectrum of applications, including medical applications, computer vision, and natural language processing. Various power-constrained platforms, such as System-on-Chips (SoCs), Application Specific Integrated Circuits (ASICs), and Field Programmable Gate Arrays (FPGAs), are explored for their utilization in implementing runtime adaptive AxC. The survey explores various techniques, such as dynamic quantization, adaptive pruning, and low-rank approximations, offering a detailed discussion of their advantages and disadvantages. Specifically, in some surveyed research works, the runtime approximation is achieved through the utilization of machine learning algorithms, with a notable emphasis on Reinforcement Learning (RL). These approaches aim to realize runtime conditions and exploit them appropriately. By providing insights into the advancements and trends in runtime adaptive AxC, this survey serves as a valuable resource for researchers and practitioners interested in this rapidly evolving area of computing. This survey conducts an in-depth investigation into the application, challenges, and scope of runtime adaptive AxC techniques, aiming to mitigate energy consumption while preserving acceptable levels of accuracy in DL models. Our primary focus lies on Convolutional Neural Networks (CNNs), with an emphasis on their application in diverse domains. In striving for comprehensiveness, the survey encompasses selected research works that extend beyond CNNs, including alternative DL models like Recurrent Neural Networks (RNNs). our scope of applications, focuses on CNNs; however, to make a comprehensive survey, we cover some surveyed research works that contain other DL models, such as RNNs. It also highlights the importance of considering specific application requirements and available resources when choosing the appropriate technique.

Artistic Neural Style Transfer (NST) has achieved remarkable success for images. However, this is not the case for dynamic 3D environments, such as computer games, where temporal coherence remains a challenge. Our paper presents an approach that uses the G-buffer information available in a game pipeline to generate robust and temporally consistent in-game artistic stylizations based on a style reference image. We use a synthetic dataset created from open-source computer games and demonstrate that the utilization of depth, normals, and edge information enables the stylization process to be more aware of the geometric and semantic aspects of a game scene. The proposed approach builds on previous work by injecting style transfer in the rendering pipeline, while also utilizing G-buffer information during inference time to improve upon the stability of the stylizations, offering a controllable way to stylize computer games in terms of temporal coherence and content preservation. Qualitative and quantitative evaluations of our in-game stylization network demonstrate significantly higher temporal stability compared to existing style transfer approaches when stylizing 3D computer games.

The emerging Processing-in-Memory (PIM) architecture shows promise in efficiently handling Deep Neural Network (DNN) inference and training by minimizing data movement through analog computing in memory. Unfortunately, the energy efficiency of PIM is still limited since it struggles to efficiently process the massive nonlinear activation functions (AFs) widely used in DNNs. Within the current solutions, the AFs require to be fulfilled in the digital domain with co-processors or lookup-table (LUT) modules. Therefore, a significant amount of data requires a round-trip conversion between the analog and the digital domains at each AF. In this way, the unavoidable analogto-digital and digital-to-analog (AD/DA) conversions dominate the system's power consumption and reduce the overall energy efficiency. To address these issues, we propose a reconfigurable analog module for nonlinear AF computation, named RAM-NAF. RAM-NAF is a pure analog module that utilizes Taylor approximation (TA) to fit the arbitrary AFs, thereby reducing AD/DA conversions. To enhance the accuracy, RAM-NAF adopts a segmentation calculation method (SCM) based on the characteristics of nonlinear AFs. Moreover, the RAM-NAF provides the capability for reconfiguration to support various AFs and can be easily integrated with the existing PIM accelerators. The experiment results show that the proposed RAM-NAF significantly improves the performance of the PIM accelerators and reduces the energy consumption: when performing inference on various PIM accelerators, the energy consumption of AD/DA conversions can be reduced up to 12.31×, while the overall energy efficiency can be increased by 2.34× to 5.20×, and the accuracy loss is below 1%.

This paper introduces a novel approach employing periodic functions for the comprehensive analysis of prime numbers. The method encompasses primality testing, factor counting and listing, prime distribution calculation, and the determination of the Nth prime. The exposition of the technique is presented in a clear and sequential manner, guiding the reader through each step with explicit equations. Graphs are strategically incorporated between crucial stages to facilitate a rapid and intuitive visualization of the rationale and outcomes of each maneuver. The paper concludes with concise reflections and ongoing inquiries into the potential applications and refinements of the proposed method.

This study introduces a hierarchical key assignment scheme (HKAS) based on the closest vector problem in an inner product space (CVP-IPS). The proposed scheme offers a comprehensive solution with scalability, flexibility, cost-effectiveness, and high performance. Key features include CVP-IPS based construction, using two public keys for the entire scheme, a distinct basis set for each class, a direct access scheme for user convenience, and rigorous mathematical and algorithmic presentation of dynamic update operations. This scheme eliminates the need for top-down structures and offers a significant benefit in that the lengths of the basis sets defined for classes are the same and the costs associated with key derivation are the same for all classes, unlike top-down approaches, where the higher class in the hierarchy incurs much higher costs. The scheme excels in both vertical and horizontal scalability due to its utilization of the access graph and is formally proven to achieve strong key indistinguishability security (S-KI-security). This research represents a significant advancement in HKAS systems, providing tangible benefits and improved security for a wide range of use cases.

The 3D point cloud plays a crucial role in deep learning-based vision tasks by providing precise spatial and depth information, leading to its increasing importance in various applications. However, the sparse nature of 3D point clouds poses computational challenges. Researchers have explored the Submanifold Sparse Convolutional Network (SSCN) for processing point cloud data while preserving sparsity. Nevertheless, existing Convolutional Neural Network (CNN) accelerators encounter difficulties in effectively handling SSCNs, prompting recent studies to focus on developing dedicated accelerators for point cloud networks to improve processing performance. This brief presents a specialized hardware architecture designed for SSCNs to address the challenges of effectively processing sparse 3D point clouds. The proposed accelerator achieves a significant 2.51× improvement in throughput density compared to previous works, highlighting its effectiveness in point cloud processing.

Background: In an era of increasing globalization and digital interaction, the demand for efficient, multilingual customer service solutions is paramount. This study introduces a novel software implementation of a conversational multilingual avatar-based interactive multifunctional AI kiosk and mobile application with computer vision, which is designed to address the challenges in delivering seamless, interactive, and inclusive customer service. The system integrates advanced technologies, including natural language processing, computer vision, and artificial intelligence, to create a user-friendly, avatar-based interactive platform. The purpose of this study is to explore the software's architecture, functionality, and effectiveness in providing a multilingual, conversational interface for diverse user demographics, particularly in kiosk and mobile application formats.Results: The implementation demonstrated robust performance in several key areas. The avatar-based interface, created using ReadyPlayerMe, delivered a highly interactive and engaging user experience, enhanced by computer vision technologies like OpenCV and Google ARCore for accurate face detection and tracking. The system effectively utilized Whisper API and Google Cloud's Speech-to-Text (STT) and Text-to-Speech (TTS) services to facilitate realtime language translation and processing. This ensured seamless communication in multiple languages. The integration with OpenAI's GPT-3.5 model, prompt-tuned to specific business data, provided contextually relevant and accurate responses, significantly enhancing the user experience. Critical issues such as privacy, data security, and real-time processing were effectively addressed, ensuring user trust and system efficiency. Additionally, the system demonstrated an ability to cater to a broad spectrum of industries, making it a versatile tool in varied business scenarios.Conclusions: The multifunctional conversational AI software effectively bridges the communication gap in customer service interactions, providing a scalable, secure, and user-friendly solution. Its ability to engage users in their native language, coupled with its interactive avatar-based interface, marks a significant advancement in the realm of digital customer service solutions. This implementation holds substantial potential for widespread adoption across various sectors, offering implications for enhancing global customer service standards, business efficiency, and user inclusivity. The success of this system sets the stage for future innovations in AI-driven customer interaction technologies.

A multiple scattering model for passive radiative transfer (RT) in vegetation that accounts for the vertical profile of the plant structure is developed, offering advancements over the commonly-used single-layer uniform scattering models prevalent in the vegetated land surface microwave remote sensing. The proposed model takes into account the complexities of the canopy morphology with vertical heterogeneity, enabling the representation of overlapping vegetation species applicable to diverse plant types and growth stages. Additionally, it serves as a valuable tool for understanding the influence of the vegetation vertical structure on the microwave brightness temperatures. The model is constructed based on high-order solutions to the RT equations, obtained through a numerical iterative approach with an efficient interpolation scheme for algorithm acceleration. This methodology facilitates the accurate distinction of the contributions to the brightness temperature from each scattering order and scattering mechanism, ensuring a comprehensive consideration of multiple scattering effects within various vegetated scenarios. The model is validated using the SMAPVEX12 L-band forest data set, encompassing a wide range of soil moisture variations. Comparisons are made between the brightness temperatures simulated by the newly developed multiple-scattering model with a continuous profile or layered profile and those obtained from a uniform single-layer model. Results demonstrate significant improvements in the multi-layered or the continuously profiled model, showing improved agreement with the measured brightness temperatures. Furthermore, the proposed model is parameterized by matching the high-order solutions to the RT equation to the widely adopted reduced order albedo-tau formalism. The resulting equivalent parameters are linked to the geometries and the electromagnetic properties of the vegetation layer, while also incorporating the effects of multiple scattering. Comparative analysis of the equivalent parameters derived from the layered model and those derived from the single-layer model reveals that the vertical heterogeneity of the vegetation structure has a notable influence on the effective scattering albedo and it yields a value more consistent with the albedo as chosen in the SMAP/SMOS inversion algorithms. Meanwhile, the impact of the vegetation vertical profile on the effective optical thickness and the effective transmissivity of the vegetation layer is weak.These insights are essential for the retrieval of soil moisture and vegetation characteristics including the plant vertical structures in microwave remote sensing.

This paper explores the intersection of Artificial Intelligence (AI) and digital storytelling in marketing, focusing on how AI-driven techniques can enhance emotional attachment and influence consumer behavior. With the rapid advancement of AI, its integration into marketing strategies has become crucial, particularly for personalizing consumer experiences and enhancing brand narratives. This study investigates AI's role in creating emotionally engaging narratives, a largely unexplored area in marketing and advertising. The research is motivated by the need to understand the dynamics between AI-driven techniques and emotional attachment in digital marketing. The paper hypothesizes a significant relationship between consumers' emotional attachment to brands, influenced by AI-driven storytelling, and their purchasing behavior. A mixed-methods research approach is employed to test this hypothesis, combining a survey with detailed interviews. The study assesses how emotional attachment, influenced by AI and storytelling, impacts consumer purchasing decisions and brand loyalty in online shopping. It also evaluates the effectiveness of AI-driven storytelling techniques in digital marketing campaigns from the perspective of online consumers. Preliminary findings suggest that while emotional attachment significantly influences consumer purchasing behavior, other factors also play a crucial role. The study reveals that AI's role in marketing is valued, but the essence of storytelling should remain grounded in human experiences. The paper concludes that the future of digital marketing lies in a harmonious blend of AI and traditional storytelling, where AI's data-driven insights complement the authenticity and emotional resonance of narratives. This research contributes to the understanding of AI's potential in enhancing emotional attachment-driven branding and online selling performance, offering insights for future strategies in digital marketing.

The IIoT network involves smart sensors, actuators, and technologies extending IoT capabilities across industrial sectors. With the rapid development in connected technology and communications in industrial applications, IIoT networks and devices are increasingly integrated into less secure physical environments. Anomaly detection in IIoT is crucial for cybersecurity. This paper proposes a novel anomaly detection model for IIoT systems, leveraging a hybrid deep learning (DL) model. The hybrid DL approach combines Gated Recurrent Units (GRU) and Convolutional Neural Networks (CNN) for anomaly detection in IoT edge computing. The proposed CNN+GRU model achieves a notable 94.94% accuracy, underscoring the importance of careful model selection for IIoT anomaly detection. The paper suggests exploring XGBoost with hybrid CNN+GRU architectures as a future direction for high accuracy in complex IIoT contexts. The Experimental results indicate a 96.41% accuracy, excelling in metrics like false alarm rate (FAR), recall, precision, and F1score. Based on these findings, we recommend future researchers consider advanced hybrid architectures and enhance efficiency using XGBoost with hybrid CNN+GRU. This approach holds promise for significant contributions to IIoT systems' security and Performance evolution.

Unsupervised ML-based approaches have emerged for driving critical decisions about training data samples to help solve challenges in many life critical applications. This paper proposes parallel and distributed computing unsupervised ML techniques to improve the execution time of different ML algorithms. Various unsupervised ML models are developed, implemented, and tested to demonstrate the efficiency, in terms of execution time and accuracy, of the serial methods as compared to the parallelized ones. We developed sequential, parallel, and distributed cloud computing unsupervised ML models based and determined the most efficient model through comparative analysis. As a case study, sequential, parallel, and distributed approaches of Simple K-Means, Minibatch K-means, and Fuzzy C-Means are investigated to study the developed models' efficiency using country datasets for multiple organizations to train and test the developed model. Parallel and distributed computing models are developed utilizing could computing architect, i.e., cloud Amazon SageMaker, to study their efficiency in the execution time and model accuracy. The results show that the proposed parallel and distributed Fuzzy C-Means outperforms the other two clustering methods in terms of execution time with 0.932ms and 0.623ms with a minimal impact on the accuracy of the developed models.

A document by Christoph Funda . Click on the document to view its contents.

In this summer project, we perform an analysis of the intermediate layer neurons primarily through the use of CLIP-dissect, a powerful model to describe neurons, and its Soft-WPMI similarity function. Specifically, we first validate the relative accuracy of CLIP-dissect when analyzing intermediate layer neurons using ground-truth labeling, before using the large pretrained model and the Soft-WPMI similarities to perform analysis of convolutional neural networks (CNNs) such as ResNet-50. We then create an interactive GUI with the visualizations of model, layer, and neuron-level analysis to allow anyone to access the intermediate neurons of these CNNs using output from CLIP-dissect and images from Broden. We allow users to directly search for specific concepts within any network so that they can better understand the inner workings of the models that increasingly define our daily life. Our website with an interactive version of this project can be found at https://dr4nx.github.io/clip-search/index.html.

This paper presents a comprehensive framework and actionable recommendations for the ethical integration of ChatGPT and other Large Language Models (LLMs) into the academic environment from a Diffusion of Innovation Theory perspective. It provides a critical analysis of the technology's potential benefits and risks in education and proposes a conceptual adoption model using critical DOI factors including the comparative benefits it offers, its compatibility with existing systems, its level of complexity, the ability to trial it, and its observability. Thorough guidelines on different assessment methods like surveys, interviews, and case studies are recommended to evaluate these factors regarding ChatGPT uptake. Additionally, strategies are discussed to promote responsible use, such as developing guidelines, assessing biases, and aligning with education policies. We call for interdisciplinary research between stakeholders, including educators, policymakers, and AI experts, to address LLM's multifaceted impacts on education.

Geographic Information System (GIS) and Remote Sensing have been considered significant in the military due to their spatiality in nature. In this study, the descriptive-analytical method was used to illustrate the applications of GIS in military operations, drawing lessons from land based military developments from selected studies. Recent military developments have seen various military institutions depending on reliable and accurate spatial mapping tools, for the purpose of Command, Control, Communication and Coordination in military operations. The study notes that high resolution satellite data and or drone technology integrated with machine learning and Artificial Intelligence (AI), have been utilized in the military for a variety of applications including cartography, terrain analysis, intelligence collection and dissemination, object recognition, safeguarding military vital installations, as well as historical construction. The integration of GIS machine learning and AI is of significance to the military planning and deployment, as the understanding of landscape is useful in determining strategic positions in the battle ground in real-time. The study recommends the need to train military personnel in geospatial techniques and ensure proper deployment, for fruitful military operations.

The recent widespread increase of the Mpox (formerly monkeypox) virus infections in the South Asian and African countries has raised concerns among medical professionals regarding the potential emergence of another pandemic in those regions. With the number of available test kits surpassing the count of positive/probable cases, there is a pressing need to develop a robust and lightweight classifier model to alleviate the burden of physical testing kits and expedite the detection process. The existing state-of-the-art primarily focuses on achieving high accuracy in modeling Mpox without considering factors such as modeling suitability, real-time inferencing, and adaptability to resource-constrained CPU-only mobile devices. In this research, we propose a novel lightweight binarized DarkNet53 model, referred to as BinaryDNet53, which is approximately ∼ 20× more computationally efficient and ∼ 2× more power-efficient than the current state-of-the-art. This model demonstrates smooth detection capabilities when deployed on small hand-held or embedded devices. Our work introduces large-margin feature learning and weighted loss calculation to enhance results, particularly on complex samples. We conduct experiments using the latest MSLD v2.0 dataset, showcasing the superiority of the proposed model over state-of-the-art models based on classification and computational metrics, including Watt power consumption, required memory, and GFLOPS.

This paper presents a robust bag encoding strategy based on an MLP mixer. The proposed approach introduces the mixing concept to MIL applications, which helps to generate robust bag encoding. The existing bag encoding strategies for MIL applications consider instances in the bag as independent. This assumption may restrict the performance of these algorithms. Therefore, this paper proposes MIL-Mixer, which utilizes the information between the instances to generate a robust bag encoding. We also extend MLP-Mixture to use classification token similar to vision transformers which diversify the encoding generation process. In this study, three benchmark MIL datasets are used to assess the performance of the proposed MIL-Mixer. In comparison with existing MIL approaches, the proposed MIL-Mixer achieves better performance.

This survey paper offers a thorough analysis of techniques and algorithms used in the identification of crime leaders within criminal networks. For each technique, the paper examines its effectiveness, limitations, potential for improvement, and future prospects. The main challenge faced by existing survey papers focusing on algorithms for identifying crime leaders and predicting crimes is effectively categorizing these algorithms. To address this limitation, this paper proposes a new methodological taxonomy that hierarchically classifies algorithms into more detailed categories and specific techniques. The paper includes empirical and experimental evaluations to rank the different techniques. The combination of the methodological taxonomy, empirical evaluations, and experimental comparisons allows for a nuanced and comprehensive understanding of the techniques and algorithms for identifying crime leaders, assisting researchers in making informed decisions. Moreover, the paper offers valuable insights into the future prospects of techniques for identifying crime leaders, emphasizing potential advancements and opportunities for further research. Here's an overview of our empirical analysis findings and experimental insights, along with the solution we've devised: (1) PageRank and Eigenvector centrality are reliable for mapping network connections, (2) Katz Centrality can effectively identify influential criminals through indirect links, stressing their significance in criminal networks, (3) current models fail to account for the specific impacts of criminal influence levels, the importance of socioeconomic context, and the dynamic nature of criminal networks and hierarchies, and (4) we propose enhancements, such as incorporating temporal dynamics and sentiment analysis to reflect the fluidity of criminal activities and relationships, which could improve the detection of key criminal figures as their roles or tactics evolve.

Modern apps require high computing resources for real-time data processing, allowing app users (AUs) to access real-time information. Edge computing (EC) provides dynamic computing resources to AUs for real-time data processing. However, ESs in specific areas can only serve a limited number of AUs due to resource and coverage constraints. Hence, the app user allocation problem (AUAP) becomes challenging in the EC environment. In this paper, a quantum-inspired differential evolution algorithm (QDE-UA) is proposed for efficient user allocation in the EC environment. The quantum vector is designed to provide a complete solution to the AUAP. The fitness function considers factors such as minimum ES required, user allocation rate (UAR), energy consumption, and load balance. Extensive simulations are performed along with hypotheses-based statistical analyses (ANOVA, Friedman test) to show the significance of the proposed QDE-UA. The results indicate that QDE-UA outperforms existing strategies with an average UAR improvement of 116.63%, a 77.35% reduction in energy consumption, and 46.22% enhancement in load balance while utilizing 13.98% fewer ESs.

The visual human perception of electromagnetic waves is limited to the so-called visible spectrum. Artificial extension of the human vision, e.g., via infrared cameras, is possible. Due to the expected low resolution at large wavelengths, however, this is rarely done in the low GHz regime. We investigate the quality of images obtained by simply turning a directive antenna across a scene. The procedure is straight forward and can be compared to how the lens in the human eye creates an image on the retina. Simulation data of an exemplary domestic indoor scenario illustrates how mono-frequency vision of WiFi signals might look like.

In the domain of electronic design automation (EDA), the significance of design space exploration (DSE) tools is increasingly recognized for their adeptness in optimizing performance, power, and area (PPA) in integrated circuits (ICs) via advanced parameter configuration. Conventional EDA exploration approaches tend to focus primarily on identifying optimal result points, frequently disregarding the nuances of the parameter space's attributes. This oversight can significantly influence the quality of the sought-after optimal PPA results. Typically, it is presumed that an ideal parameter space usually manifests uniform distribution across each parameter. However, this uniformity in the parameter space does not necessarily assure the discovery of superior pareto points. To address this, our study introduces a novel method called Shuffled Iterative Hierarchical Minimixing (SIHM) that it can incorporate the reduced sampling within a uniformly constructed parameter space, supplemented by mixed superposition with noise. This strategy could notably enhance the quality of pareto points within the PPA results. Our experimental findings reveal that this approach has the ability to yield improved pareto points with a data sampling ratio as low as approximately 1:96 during each parameter, while maintaining exceptionally low noise levels. Additionally, our pareto points show a 3.9% decrease at least in the minimum value within one of the dimensional space in the PPA results, compared to pareto points derived from uniformly distributed parameter space.

Introduction: The FAIR principles stand for four important features that research objects such as software, methods and data should follow in order to allow other researchers replicate the analysis and the results. The four components of these principles are the findability, accessibility, interoperability and reusability, and most the software available and associated to publications are findable by versioning systems such as github, but often they do not meet the other criteria, either because there are missing data, lack of documentation about the usage or testable examples to guarantee the appropriate execution. Regarding the usability, the metadata describing the objects allow the correct categorization of the research objects and proper recommendation of its application. But the metadata generation using semantic standards requires some knowledge of controlled vocabularies to describe them correctly, allowing an exclusive portion of the community to fully meet the FAIR principles. Results: We implemented the FAIRLibJS that provides means to annotate Javascript software libraries allowing serverless web applications achieve the fourth FAIR criteria through correct metadata generation following semantic metadata standards. Besides the software metadata generation, the tool also enables the search and automatic semantic description of scientific publications indexed in OpenAire. The third main functionality is the extraction of keywords for any web application software and, for those directed to the biological domain, it also provides a mapping with the EDAM ontology topic and operation concepts. The JavaScript library and the Web module with usage examples and functionality demonstrations is provided to the public domain with open source. Conclusion: The FAIRLibJS was able to extract correct keywords, from a given tool summary, for the manually annotated workflows in workflowHub, and also chose correctly the topics and operations concepts for more than 90% of the workflows. We also showed that through the usage of a few extra commentary syntax for documentation strings, the software developers are able to use our tool to derive enriched metadata to describe the Javascript modules, their functions with inputs and returned outputs and the module dependencies. The tool also provides a friendly interface to search data in OpenAire platform, enabling the automatic annotation of scientific publications.

Pyramid Temporal Hierarchy Network (PTH-Net) is a new paradigm for dynamic facial expression recognition, applied directly to raw videos without face detection and alignment (FDA). The traditional paradigm initially employs FDA to extract facial regions from raw videos before recognition. The advantage of this paradigm lies in minimizing the impact of complex backgrounds. However, it inadvertently neglects valuable information, such as body movements. Additionally, being bound to FDA sacrifices flexibility. In contrast, PTH-Net distinguishes background and target at the feature level, preserves more critical information, and is an end-to-end network that is more flexible. Specifically, PTH-Net utilizes a pre-trained backbone to extract multiple generic features of video understanding at various temporal frequencies, forming pyramid features. Subsequently, through temporal hierarchy refinement—achieved via differential sharing and downsampling—PTH-Net refines key information under the supervision of multiple receptive fields with the temporal-frequency invariance of expressions. In addition, to solve the problem of containing numerous irrelevant frames in videos, PTH-Net incorporates a Temporal Hierarchy Refinement layer to aggregate information at different temporal granularities, enhancing its ability to distinguish target and non-target expressions. Notably, PTH-Net achieves more comprehensive and in-depth understanding by merging knowledge from both forward and reverse video sequences. PTH-Net excels across six challenging benchmarks with lower computational costs in comparison to preceding methods.