Network Selection Research Articles

One for Dozens: Adaptive REcommendation for All Domains with Counterfactual Augmentation

Multi-domain recommendation (MDR) aims to enhance recommendation performance across various domains. However, real-world recommender systems in online platforms often need to handle dozens or even hundreds of domains, far exceeding the capabilities of traditional MDR algorithms, which typically focus on fewer than five domains. Key challenges include a substantial increase in parameter count, high maintenance costs, and intricate knowledge transfer patterns across domains. Furthermore, minor domains often suffer from data sparsity, leading to inadequate training in classical methods. To address these issues, we propose Adaptive REcommendation for All Domains with counterfactual augmentation (AREAD). AREAD employs a hierarchical structure with a limited number of expert networks at several layers, to effectively capture domain knowledge at different granularities. To adaptively capture the knowledge transfer pattern across domains, we generate and iteratively prune a hierarchical expert network selection mask for each domain during training. Additionally, counterfactual assumptions are used to augment data in minor domains, supporting their iterative mask pruning. Our experiments on two public datasets, each encompassing over twenty domains, demonstrate AREAD's effectiveness, especially in data-sparse domains.

Deep Learning-Based Beam Selection in RIS-Aided Maritime Next-Generation Networks with Application in Autonomous Vessel Mooring

Deep Learning-Based Beam Selection in RIS-Aided Maritime Next-Generation Networks with Application in Autonomous Vessel Mooring

Transfer learning based feature selection for feedforward neural network for speech emotion classifier

This work discusses speech emotion recognition via custom feature engineering and feature selection techniques using mel-frequency cepstral coefficients as initial audio features. Proposed transfer learning approach consist in employing the backward-step selection algorithm for feature selection using statistical learning classifiers, the obtained subset of features than subsequently used to train feedforward neural networks. This technique allowed us to significantly reduce initial feature vector size while increasing models’ prediction quality. We used TESS and RAVDESS datasets to estimate the performance of proposed method. To evaluate the quality of the model, unweighted average recall (UAR) was used. Experimental results demonstrate promising accuracy (UAR = 82 % for TESS and UAR = 53 % for RAVDESS), showcasing the potential of this approach for applications like virtual agents, voice assistants and mental health diagnostics.

Spectral efficient network and resource selection model in 5G networks

This work addresses the challenges in modern communication networks, emphasizing the need for improved efficiency, higher data transfer rates, and reduced delays. In 5G networks, advanced resource optimization, network selection, and relaying techniques are crucial for expanding multi-cellular coverage and enhancing network performance. However, implementing these techniques in mobile environments with high interference levels increases computational demands for radio resource management (RRM). Machine learning (ML) and deep learning (DL) are proposed as solutions to enhance consumer applications, reduce communication overhead, and improve RRM. Current ML/DL methods, however, struggle with identifying key features for network selection and balancing system throughput with spectral efficiency. This paper introduces the spectral efficient network and resource selection (SENRS) model for 5G multiple input multiple output-orthogonal frequency division multiplexing (MIMO-OFDM) networks. Tested using the Stanford University Interim (SUI) channel fading model in a highway scenario, the SENRS model demonstrates superior performance compared to existing network and resource selection systems.

Online Learning-Based Joint Gateway Selection and User Scheduling in Non-Stationary Air-Ground Networks

Online Learning-Based Joint Gateway Selection and User Scheduling in Non-Stationary Air-Ground Networks

Priority Based Reliable Adaptive Relay Selection and Cooperative Multi-Hop Clustering for Vehicular Networks

Priority Based Reliable Adaptive Relay Selection and Cooperative Multi-Hop Clustering for Vehicular Networks

A standard network selection and resource allocation mechanism in 5G heterogeneous networks using hybrid heuristic algorithm with multi-objective constraints

The integration of various Random Access Technologies (RATs) within 5G Heterogeneous Networks (HetNets) to satisfy the diverse communication needs of the Internet of Things (IoT) causes significant challenges in network topology selection and resource allocation. Traditional approaches do not handle network congestion and poor user experience which necessitates the development of more efficient and intelligent network management strategies. The main novelty of the research work is to combine the two intelligent optimization algorithms to address the complexities of resource management for enhancing system performance. The integrated optimization algorithm also aims to reduce the latency and communication costs while enhancing resource utilization within the network. The novel Hybrid Snow Leopard and Dark Forest Algorithm (HSL-DFA) combines the strengths of the Snow Leopard Optimization Algorithm (SLOA) and the Dark Forest Algorithm (DFA) to optimize network performance based on the multiple objectives including resource utilization, makespan, Quality of Service (QoS), energy consumption, communication cost, congestion control, and latency. The HSL-DFA algorithm integrates the SLOA and DFA to leverage their respective advantages in solving complex optimization problems. It worked based on a position-updating process using current and mean fitness values. Here, the SLOA is used for the position-updating process if the current fitness exceeds the mean fitness and DFA is used for the opposite scenario. This approach ensures higher convergence rates and optimal solutions for complex network optimization problems. By addressing multi-objective constraints, the algorithm significantly improves network performance and provides a promising solution for the management of a 5G network. Various metrics are utilized to confirm the effectiveness of the proposed model. The results showed that the throughput of the proposed model was 93 at the 200th node.

Multiobjective Fractional Transit Waterwheel Optimization for Energy Efficient CH Selection and Routing

ABSTRACTIn this research, an effective approach named Fractional Transit Water Wheel Optimization (FTWWO) is introduced for cluster head (CH) selection and routing in wireless sensor networks (WSNs) with mobile sink. Initially, WSN simulation with mobile sink is carried out based on the energy model, mobility model, and link life time (LLT) model. Then, the cluster formation is done by deep embedding clustering (DEC). After that, the CH selection is carried out by using a novel Hybrid Transit Water Wheel Optimization (TWWO) method, based on factors like predicted energy, intracluster and intercluster distances, LLT, trust, and delay. Moreover, the TWWO algorithm is designed by integrating the Transit Search Algorithm (TSA) and the Water Wheel Plant Algorithm (WWPA). Following this, routing is carried out by using a proposed fractional TWWO, by considering the factors that include predicted energy, distance, delay, and trust, in which the energy prediction is performed by the radial basis function networks (RBFNs). The fractional TWWO is developed by combining fractional calculus (FC) with the proposed TWWO. Finally, to enhance network lifespan, energy‐efficient data aggregation techniques are applied. Additionally, the proposed method gained a minimum distance of 39.332 m, energy of 0.001 J, latency of 0.547 s, maximum throughput of 35.106 Mbps, and packet delivery rates (PDR) of 81.637%.

DCASR: Distributed Collaborative Authentication With Specified Security Strength and Resource Optimization Selection in AAV Networks

DCASR: Distributed Collaborative Authentication With Specified Security Strength and Resource Optimization Selection in AAV Networks

Low-Complexity Relay Selection for Full-Duplex Random Relay Networks

Full-duplex relay networks have been studied to enhance network performance under the assumption that the number and positions of relay nodes are fixed. To account for the practical randomness in the number and locations of relays, this paper investigates full-duplex random relay networks (FDRRNs) where all nodes are randomly distributed following a Poisson point process (PPP) model. In addition, we propose a low-complexity relay selection algorithm that constructs the candidate relay set while considering the selection diversity gain. Our simulation results demonstrate that, rather than simply increasing the number of candidate relay nodes, selecting an appropriate candidate relay set can achieve significant performance enhancement without unnecessarily increasing system complexity.

Q-learning based heterogeneous network selection decision algorithm for ultra reliable and low latency communication services

Q-learning based heterogeneous network selection decision algorithm for ultra reliable and low latency communication services

Towards the automatic detection of activities of daily living using eye-movement and accelerometer data with neural networks.

Towards the automatic detection of activities of daily living using eye-movement and accelerometer data with neural networks.

Joint Deep Reinforcement Learning and Unfolding for Sensing and Communication Function Selection in Vehicular Networks

Joint Deep Reinforcement Learning and Unfolding for Sensing and Communication Function Selection in Vehicular Networks

UAV path selection in multi-hop cooperative UAV-based networks: A deep reinforcement learning approach

UAV path selection in multi-hop cooperative UAV-based networks: A deep reinforcement learning approach

Neurocognitive and resting-state functional MRI changes in patients with diffuse gliomas after chemoradiotherapy.

Neurocognitive and resting-state functional MRI changes in patients with diffuse gliomas after chemoradiotherapy.

Quantum neural network-based approach for optimizing road network selection

With advancements in neural networks, intelligent road network selection methods have become a key research area. However, the expanding scale of road networks has led to concerns regarding model training efficiency and resource consumption. Quantum neural networks, leveraging their unique properties of superposition and entanglement, present remarkable advantages for handling large-scale, complex, and nonlinear data. We propose a novel framework for road network selection based on quantum neural networks. We design a comprehensive feature set that accounts for various factors, including terrain, settlements, and surrounding density. Our study delves into the impact of feature encoding methods and circuit structures on the performance of quantum neural networks in road selection. It evaluates the proposed model’s performance across different scales, regions, and data volumes. The results demonstrate the feasibility and effectiveness of our approach when compared to existing classical neural network models, offering a promising solution for large-scale road network selection.

It's who you know: a review of peer networks and academic achievement in schools.

Cultivating equitable academic achievement and enhancing academic functioning for youth is widely seen as one of the main missions of the public education system in the US, but educators may be overlooking critical factors for fully realizing this collective mission. This paper explores the emergent evidence on the impact of friendship and peer social networks on academic achievement and outcomes for students in K-12 public schools in the United States. In total, nine studies have been reviewed and presented. Findings suggest that peer social network selection and influence effects impact individual and group academic functioning and achievement outcomes across age groups in adolescence. Evidence across multiple studies suggests that structural effects and network position may be of special importance, particularly for low-achieving youth. However, due to publication bias issues and methodological limitations, findings should be taken cautiously. Social Network Analysis (SNA) is a quickly growing subfield of education research. There is a great need for more studies in the US, in particular those focusing on historically marginalized and underrepresented student populations (e.g., BIPOC and LGBTQ+ populations, highly mobile student groups, foster youth, students qualifying for special education support and accommodation, etc.). Another major gap identified is the need to focus on specific developmental stages, like elementary school or early adolescence, and varying environmental contexts.

Automated Energy Performance Monitoring and Occupant Engagement in Buildings through Blockchain-Enabled Non-Fungible Tokens

Building energy efficiency programs face significant challenges in performance monitoring and occupant engagement, which hinder the achievement of sustainability goals in the built environment. Traditional systems often suffer from intermediary-dependent workflows, insufficient transparency, and reliability issues, leading to conflicts among stakeholders and reduced occupant participation. This study proposes a blockchain-enabled solution that leverages Non-Fungible Tokens (NFTs) to improve the transparency, reliability, and traceability of performance monitoring systems. By integrating Digital Twin (DT) technology, blockchain, and a token marketplace, the platform not only enhances monitoring capabilities but also incentivizes occupants to adopt energy-efficient behaviors through Fungible Token (FT) rewards. A proof-of-concept prototype was developed using a synthetic case study, demonstrating the feasibility, cost efficiency, and scalability of the framework. The findings emphasize the importance of network selection for wider blockchain adoption. This transparent and immutable framework addresses key challenges in energy performance monitoring, offering a foundation for advancing sustainability in the built environment.

Injecting complexity in simulation models: Do selection and social influence jointly promote cooperation?

This paper employs a simulation model to investigate the effectiveness of cooperation selection (“selecting similar others”) and social influence (“do as others do”), since both mechanisms promote cooperation in theoretical analyses and experimental studies. However, it is unclear how effective cooperation selection and social influence are in simulation models where both mechanisms operate simultaneously alongside additional social dynamics, such as reciprocity and transitivity. This paper relies on a model loosely based on an empirical case in which students selected others based on how cooperative they perceive the other. Using existing theoretical cooperation models as a benchmark, we insert relational, behavioral, and contextual assumptions into our model and build on data from 95 students when we vary the strength of cooperation selection and social influence relative to empirically observed levels. We take co-evolution stochastic actor-oriented models as basis because the model inherently accounts for the interdependence of behavior and network selection. Our simulations reveal that cooperation benefits most when cooperation selection and social influence are strongly positive. Through the combination of cooperation selection and social influence, the simulations show that cooperators form dense local clusters, influencing their peers to keep cooperating while insulating themselves from social influence from defectors. Robustness checks confirm the stability of these findings across diverse parameter configurations.

KAT7-acetylated YBX1 promotes hepatocellular carcinoma proliferation by reprogramming nucleotide metabolism

BackgroundLysine acetylation is a critical post-translational modification regulating tumor initiation and progression. Lysine acetyltransferase 7 (KAT7)-mediated lysine acetylation is frequently dysregulated in cancer. However, the role of KAT7-mediated lysine acetylation in hepatocellular carcinoma (HCC) progression remains unclear.MethodsBioinformatic analysis was used to investigate the expression, clinicopathological characteristics and diagnostic prognostic value of KAT7 in HCC. CCK-8 assays, colony-forming assays, apoptosis assays and nude mouse xenograft models were utilized to detect the oncogenic functions of KAT7 in HCC. Immunoprecipitation (IP) assay and mass spectrometry (MS) analysis were performed to identify the KAT7-binding protein Y-box binding protein 1 (YBX1). Transcriptome sequencing and functional enrichment analysis were employed to elucidate the downstream pathway regulated by KAT7 and YBX1. Chromatin immunoprecipitation (ChIP) assay was used to evaluate YBX1 binding to the promoter regions of ribonucleotide reductase regulatory subunit M2 (RRM2) and thymidine kinase 1 (TK1). Weighted gene co-expression network analysis and selection operator regression analysis were used to build risk prediction models.ResultsThis study demonstrated that elevated KAT7 expression is associated with poor prognosis in HCC patients. Knockdown of endogenous KAT7 in HCC cells attenuated tumorigenic phenotypes associated with cell proliferation, colony formation and orthotopic xenograft tumor growth, indicating a pro-tumorigenic role of KAT7 in HCC. YBX1 was identified as a novel non-histone substrate for KAT7, and the E508 residue of KAT7 is essential for binding. Following the functional enrichment analysis, KAT7 and YBX1 were correlated with nucleotide metabolism. Furthermore, KAT7 binds to YBX1 and modulates its post-translational expression, which enhances the transcriptional activity of the central nucleotide metabolism enzymes RRM2 and TK1. Additionally, we constructed a novel prognostic prediction model based on KAT7, YBX1, RRM2 and TK1, which validated the predictive accuracy and prognostic value of KAT7-mediated acetylation is consistent with clinical outcomes in HCC.ConclusionsOur findings highlight that KAT7 acetylates YBX1 and promotes HCC progression by reprogramming nucleotide metabolism, offering therapeutic implications.