Djamel Bouchaffra, Faycal Ykhlef, Hanene Azzag
et al.
Standard attention mechanisms in transformers are limited by their pairwise formulation, which hinders the modeling of higher-order dependencies among tokens. We introduce the NeuroGame Transformer (NGT) to overcome this by reconceptualizing attention through a dual perspective: tokens are treated simultaneously as players in a cooperative game and as interacting spins in a statistical physics system. Token importance is quantified using two complementary game-theoretic concepts -- Shapley values for global, permutation-based attribution and Banzhaf indices for local, coalition-level influence. These are combined via a learnable gating parameter to form an external magnetic field, while pairwise interaction potentials capture synergistic relationships. The system's energy follows an Ising Hamiltonian, with attention weights emerging as marginal probabilities under the Gibbs distribution, efficiently computed via mean-field equations. To ensure scalability despite the exponential coalition space, we develop importance-weighted Monte Carlo estimators with Gibbs-distributed weights. This approach avoids explicit exponential factors, ensuring numerical stability for long sequences. We provide theoretical convergence guarantees and characterize the fairness-sensitivity trade-off governed by the interpolation parameter. Experimental results demonstrate that the NeuroGame Transformer achieves strong performance across SNLI, and MNLI-matched, outperforming some major efficient transformer baselines. On SNLI, it attains a test accuracy of 86.4\% (with a peak validation accuracy of 86.6\%), surpassing ALBERT-Base and remaining highly competitive with RoBERTa-Base. Code is available at https://github.com/dbouchaffra/NeuroGame-Transformer.
Juan Fernando Lima, María Inés Acosta-Urigüen, Marcos Orellana
In psychology, clinical instruments are often used for assessing the cognitive domains and detecting mental alterations. Nevertheless, after the screening stage, the recommendation of activities for training or new assessments could take a long time. Other instruments such digital serious games are commonly used to train cognitive domains, unlike classical instruments, can collect extra data on players' behavior without altering a person's state. In the other hand, decision-making tools are based mainly on two techniques, machine learning for revealing hidden data patterns, and knowledge engineering for building knowledge-based systems. However, the synergy among these both techniques is not completely leveraged, otherwise, when it is, this intelligent data analysis allows the creation of hypotheses and validate contradictions. In this context, this paper presents a tool for supporting the cognitive memory assessment of people based on machine learning and knowledge engineering (M&K-CogMem). The source of data is a matching game for memory training, and after that, this tool determines the memory status of players among their respective age groups, it based on the scores and times in the game. In case of detecting a cognitive memory alteration, it recommends activities for memory improvement through its inference engine. Testing its adoption in industry, an empirical evaluation carried out by psychologists through a technology acceptance model showed positive indicators.
Objectives. Analysis of the various architectures of computing systems (CSs) used in recent decades has allowed us to identify the most common structures. One of the key features is the use of mass-produced equipment to create data processing subsystems (for example, multicore processors and high-capacity semiconductor memory), as well as network equipment to build communication subsystems. This reduces hardware costs and allows typical or cluster configurations to be created, which is especially important for expensive CSs. The desire to achieve high computational speed and performance in such CSs requires minimizing the time to complete the task and balancing time delays both in data processing subsystems and in the communication subsystem which provides data transmission inside the CS. The aim of this work is to analyze computing modules (CMs) and structures on the basis of which the construction of cluster CSs is carried out.Methods. The main results of the work were obtained using methods of mathematical analysis and modeling.Results. The study considers the structure of modern multicore microprocessors as the basis for building CMs of cluster CSs. As the number of cores in the microprocessor structure increases, the communication network which unites them into a single structure becomes more complicated. It has been shown that in new developments of microprocessors, communication between cores is performed in the form of a network. The microprocessors themselves are MIMD structures in accordance with the well-known Flynn classification.Conclusions. The proposed method of selecting an effective structure of a CS allows us to obtain the optimal structure of a CS according to the criterion of performance.
Palanichamy Naveen, Rajagopal Maheswar, Pavel Trojovský
Integrating natural language processing (NLP) techniques with spatial data infrastructures (SDIs) potentially revolutionize the way users interact with geospatial data. This article presents GeoNLU, a comprehensive framework aimed at bridging the gap between natural language and SDIs. GeoNLU aims to enable seamless interaction and querying of geospatial data through natural language, thereby enhancing accessibility and usability for a wide range of users. This article delves into the theoretical foundations, architectural design, key components, and potential applications of GeoNLU, highlighting its significance in improving geospatial data exploration, analysis, and decision-making.
Background: The Fourth Industrial Revolution (4IR) has transformed value chains across various industries. However, there is still a lack of knowledge on how to effectively utilise 4IR technologies in organisational value chains. Research must explore how organisations can use 4IR technologies to optimise their value chain performance.
Objectives: The primary aim of this study is to explore the holistic impact of 4IR technologies on the structural transformation of value chains.
Method: Applying the principles of the resource-based view and the VRIO (Valuable, Rare, Inimitable and Organisation) framework, a systematic literature review was conducted to map the intersection of 4IR technologies and value chain performance. It utilised resource-based view theory and the VRIO framework to assess the role of 4IR in transforming value chains. This study focused on the VRIO integration of 4IR resources, such as advanced data analytics, Internet of Things (IoT), proprietary technologies and skilled workforces.
Results: The research shows that 4IR resources are valuable and rare assets that require meticulous organisational integration, adaptable organisational structures, innovation-driven cultures and cross-functional collaboration.
Conclusion: The strategic integration of 4IR resources within value chains can lead to innovation, efficiency and enduring competitive advantage.
Contribution: This study provides a strategic roadmap for integrating emerging 4IR technologies into business value chains, cultivating a deeper understanding and maximising the benefits of these technologies to achieve sustained value creation and competitive advantage.
Management information systems, Information theory
Since the introduction of Generative Adversarial Networks (GANs) in speech synthesis, remarkable achievements have been attained. In a thorough exploration of vocoders, it has been discovered that audio waveforms can be generated at speeds exceeding real-time while maintaining high fidelity, achieved through the utilization of GAN-based models. Typically, the inputs to the vocoder consist of band-limited spectral information, which inevitably sacrifices high-frequency details. To address this, we adopt the full-band Mel spectrogram information as input, aiming to provide the vocoder with the most comprehensive information possible. However, previous studies have revealed that the use of full-band spectral information as input can result in the issue of over-smoothing, compromising the naturalness of the synthesized speech. To tackle this challenge, we propose VNet, a GAN-based neural vocoder network that incorporates full-band spectral information and introduces a Multi-Tier Discriminator (MTD) comprising multiple sub-discriminators to generate high-resolution signals. Additionally, we introduce an asymptotically constrained method that modifies the adversarial loss of the generator and discriminator, enhancing the stability of the training process. Through rigorous experiments, we demonstrate that the VNet model is capable of generating high-fidelity speech and significantly improving the performance of the vocoder.
Link prediction typically studies the probability of future interconnection among nodes with the observation in a single social network. More often than not, real scenario is presented as a multiplex network with common (anchor) users active in multiple social networks. In the literature, most existing works study either the intra-link prediction in a single network or inter-link prediction among networks (a.k.a. network alignment), and consider two learning tasks are independent from each other, which is still away from the fact. On the representation space, the vast majority of existing methods are built upon the traditional Euclidean space, unaware of the inherent geometry of social networks. The third issue is on the scarce anchor users. Annotating anchor users is laborious and expensive, and thus it is impractical to work with quantities of anchor users. Herein, in light of the issues above, we propose to study a challenging yet practical problem of Geometry-aware Collective Link Prediction across Multiplex Network. To address this problem, we present a novel contrastive model, RCoCo, which collaborates intra- and inter-network behaviors in Riemannian spaces. In RCoCo, we design a curvature-aware graph attention network ($κ-$GAT), conducting attention mechanism in Riemannian manifold whose curvature is estimated by the Ricci curvatures over the network. Thereafter, we formulate intra- and inter-contrastive loss in the manifolds, in which we augment graphs by exploring the high-order structure of community and information transfer on anchor users. Finally, we conduct extensive experiments with 14 strong baselines on 8 real-world datasets, and show the effectiveness of RCoCo.
Self-supervised learning has demonstrated impressive performance in speech tasks, yet there remains ample opportunity for advancement in the realm of speech enhancement research. In addressing speech tasks, confining the attention mechanism solely to the temporal dimension poses limitations in effectively focusing on critical speech features. Considering the aforementioned issues, our study introduces a novel speech enhancement framework, HFSDA, which skillfully integrates heterogeneous spatial features and incorporates a dual-dimension attention mechanism to significantly enhance speech clarity and quality in noisy environments. By leveraging self-supervised learning embeddings in tandem with Short-Time Fourier Transform (STFT) spectrogram features, our model excels at capturing both high-level semantic information and detailed spectral data, enabling a more thorough analysis and refinement of speech signals. Furthermore, we employ the innovative Omni-dimensional Dynamic Convolution (ODConv) technology within the spectrogram input branch, enabling enhanced extraction and integration of crucial information across multiple dimensions. Additionally, we refine the Conformer model by enhancing its feature extraction capabilities not only in the temporal dimension but also across the spectral domain. Extensive experiments on the VCTK-DEMAND dataset show that HFSDA is comparable to existing state-of-the-art models, confirming the validity of our approach.
Speech self-supervised learning (SSL) represents has achieved state-of-the-art (SOTA) performance in multiple downstream tasks. However, its application in speech enhancement (SE) tasks remains immature, offering opportunities for improvement. In this study, we introduce a novel cross-domain feature fusion and multi-attention speech enhancement network, termed BSS-CFFMA, which leverages self-supervised embeddings. BSS-CFFMA comprises a multi-scale cross-domain feature fusion (MSCFF) block and a residual hybrid multi-attention (RHMA) block. The MSCFF block effectively integrates cross-domain features, facilitating the extraction of rich acoustic information. The RHMA block, serving as the primary enhancement module, utilizes three distinct attention modules to capture diverse attention representations and estimate high-quality speech signals. We evaluate the performance of the BSS-CFFMA model through comparative and ablation studies on the VoiceBank-DEMAND dataset, achieving SOTA results. Furthermore, we select three types of data from the WHAMR! dataset, a collection specifically designed for speech enhancement tasks, to assess the capabilities of BSS-CFFMA in tasks such as denoising only, dereverberation only, and simultaneous denoising and dereverberation. This study marks the first attempt to explore the effectiveness of self-supervised embedding-based speech enhancement methods in complex tasks encompassing dereverberation and simultaneous denoising and dereverberation. The demo implementation of BSS-CFFMA is available online\footnote[2]{https://github.com/AlimMat/BSS-CFFMA. \label{s1}}.
Nils Mandischer, Marcel Usai, Frank Flemisch
et al.
The implicit assumption that human and autonomous agents have certain capabilities is omnipresent in modern teaming concepts. However, none formalize these capabilities in a flexible and quantifiable way. In this paper, we propose Capability Deltas, which establish a quantifiable source to craft autonomous assistance systems in which one agent takes the leader and the other the supporter role. We deduct the quantification of human capabilities based on an established assessment and documentation procedure from occupational inclusion of people with disabilities. This allows us to quantify the delta, or gap, between a team's current capability and a requirement established by a work process. The concept is then extended to the multi-dimensional capability space, which then allows to formalize compensation behavior and assess required actions by the autonomous agent.
Hanxian He, Campbell Wilson, Thanh Thi Nguyen
et al.
When it comes to classifying child sexual abuse images, managing similar inter-class correlations and diverse intra-class correlations poses a significant challenge. Vision transformer models, unlike conventional deep convolutional network models, leverage a self-attention mechanism to capture global interactions among contextual local elements. This allows them to navigate through image patches effectively, avoiding incorrect correlations and reducing ambiguity in attention maps, thus proving their efficacy in computer vision tasks. Rather than directly analyzing child sexual abuse data, we constructed two datasets: one comprising clean and pornographic images and another with three classes, which additionally include images indicative of pornography, sourced from Reddit and Google Open Images data. In our experiments, we also employ an adult content image benchmark dataset. These datasets served as a basis for assessing the performance of vision transformer models in pornographic image classification. In our study, we conducted a comparative analysis between various popular vision transformer models and traditional pre-trained ResNet models. Furthermore, we compared them with established methods for sensitive image detection such as attention and metric learning based CNN and Bumble. The findings demonstrated that vision transformer networks surpassed the benchmark pre-trained models, showcasing their superior classification and detection capabilities in this task.
A novel proposed Binary Volleyball Premier League algorithm (BVPL) has shown some promising results in a Parkinson’s Disease (PD) dataset related to fitness and accuracy [1]. This paper evaluates and provides an overview of the efficiency of BVPL in feature selection compared to various metaheuristic optimization algorithms and PD datasets. Moreover, an improved variant of BVPL is proposed that integrates the opposite-based solution to enlarge search domains and increase the possibility of getting rid of the local optima. The performance of BVPL is validated using the accuracy of the k-Nearest Neighbor Algorithm. The superiority of BVPL over the competing algorithms for each dataset is measured using statistical tests. The conclusive results indicate that the BVPL exhibits significant competitiveness compared to most metaheuristic algorithms, thereby establishing its potential for accurate prediction of PD. Overall, BVPL shows high potential to be employed in feature selection.
Kosovo’s private sector business is led by small and medium enterprises and shows a crucial role in the economic development and promoting the growth of this sector. Due to the lack of studies in Kosova, related to marketing strategy and export performance, the research aims to explore growth determinants of SMEs in Kosovo by emphasizing exports, revealing the importance of marketing strategy related to the performance of SMEs. Since exporting is becoming more important, it is necessary to understand all the factors that are involved in this booming trade. This research analyses explore the literature dealing with small and medium enterprises' performance to inspect the traditional and modern academic standpoints on estimating the impact of export performance, based on secondary data of five hundred small and medium enterprises, as trade, manufacturing, and services sectors in Kosovo. The data were analysed using both descriptive statistics and inferential techniques, such as logistic regression analysis, and probit model using the SPSS software. Results show a significant linkage of marketing strategy and export performance; managers and policymakers can take some action to improve export performance. On the other hand, the results measure the impact of price, production, promotion and place in export performance incorporate expected future performance. It is argued that enterprises in Kosovo must pay more attention to strategies, and attempt to design and implement better strategies that will be beneficial for developing international trade, especially exports, as one way of trading by SMEs.
A fuzzy C-Means segmentation algorithm can be implemented in an image segmentation
based on the Mahalanobis distance; However, this method only needs to consider the color
space situation, not the neighborhood system of the image. It was an effective edge detection
process unwell performed and generated less accuracy in segmentation results. In this article,
we propose a new method for image segmentation with Mahalanobis fuzzy C-means Spatial
information (MFCMS). The proposed method combines feature space and images of the
information of the neighborhood (spatial information) to improve the accuracy of the result of
segmentation on the image. The MFCMS consists of two steps, the histogram threshold module
for the first step and the MFCMS module for the second step. The Histogram Threshold module
is used to get the MFCMS initialization conditions for the cluster centroid and the number of
centroids. Test results show that this method provides better segmentation performance than
classification errors (ME) and relative foreground area errors (RAE) of 1.61 and 3.48,
respectively.
Presently, the grid-connected scale from photovoltaic (PV) system is getting higher among renewable power generations. However, the PV output power can be affected by different meteorological conditions due to PV randomness and volatility. Accordingly, reasonable generation plans can be well arranged using accurate PV power prediction among various types of energy sources, thus reducing the effect of PV system on the grid. To resolve this problem, a PV output power prediction model, namely IMWOASVM, is proposed based on the combination of improved whale optimization algorithm (IMWOA) and support vector machine (SVM). The IMWOA is used to optimize the kernel function parameter and penalty coefficient in SVM. The optimal parameter and coefficient values can then be input to SVM for enhancing the PV prediction. The performance results verify that the coefficient of determination using the IMWOA model can reach beyond 99% in both sunny and cloudy days. Simultaneously, the mean absolute errors on sunny and cloudy days are 0.0251 and 0.0705, respectively. The root mean square errors in sunny and cloudy days are 2.17% and 1.03%, respectively. The results confirm that the proposed model effectively increases the accuracy of the PV output power prediction and is superior to existing methods.
Viktor S. Gorbatov, Igor Y. Zhukov, Vladislav V. Kravchenko
et al.
The purpose of this paper is an analytical pre-project study of possible technological aspects of countering external computer attacks on critical network infrastructure. This will make it possible to specify the tasks for further resolving this problem in the aspect of developing the necessary software and hardware. The practical implementation of such tasks is an urgent and rather unconventional problem due to various factors of change in the classical concept of the network perimeter as a physical boundary of the information infrastructure, which becomes virtual and, therefore, requires the use of new approaches to the development of technical solutions. Based on statistical data on the number and quality of computer incidents, the study provides a justification for the relevance of the above problem, and gives an overview of widely used technical means for protecting the classic network perimeter, such as firewalls and systems for detecting attacks and intrusions. A comparative analysis of modern technological trends in their development, referred to in publications as «Threat Detection and Response», «Extended Detection and Response», is carried out. However, despite the powerful software and hardware functionality of these solutions, their common drawback is indicated as the lack of adequate counteraction to computer attacks with a remote mode of the user work. In this regard, the latest concept of virtual network perimeter protection, referred to by the authors as «Cybersecurity Mesh» («cybersecurity network»), is detailed. It is this methodology that seems to be the most promising for the development of appropriate technological solutions to ensure the cybersecurity of the perimeter of the critical information infrastructure. The paper might be useful to specialists working on the security of critical information infrastructure facilities, as well as to employees of educational classes in the implementation of appropriate training, retraining and advanced training programs for such specialists.