Individual Edges Research Articles

Task Offloading in Multi-Hop Relay-Aided Multi-Access Edge Computing

As demands for multi-access edge computing (MEC) increase exponentially, resource limitations at individual edge servers (ESs) will inevitably become bottlenecks. Most existing works largely ignore spot access spectrum resource availability when considering computing cooperation among ESs. In this paper, we present a general cooperative MEC framework to enable network-wide cooperation of ESs by leveraging a multi-hop task relaying strategy. Under this framework, we can maximize overall throughput (task completion rate) by balancing the workload for both spot communication and computing through task routing. To address the problem, we formulate an optimization problem and apply matching theory to find a task offloading scheme with low-complexity. Finally, we conduct numerical analysis to show the effectiveness of our proposed scheme and the superiority of our multi-hop relay-assisted MEC systems.

TransMUSE: Transferable Traffic Prediction in MUlti-Service Edge Networks

The Covid-19 pandemic has forced the workforce to switch to working from home, which has put significant burdens on the management of broadband networks and called for intelligent service-by-service resource optimization at the network edge. In this context, network traffic prediction is crucial for operators to provide reliable connectivity across large geographic regions. Although recent advances in neural network design have demonstrated potential to effectively tackle forecasting, in this work we reveal based on real-world measurements that network traffic across different regions differs widely. As a result, models trained on historical traffic data observed in one region can hardly serve in making accurate predictions in other areas. Training bespoke models for different regions is tempting, but that approach bears significant measurement overhead, is computationally expensive, and does not scale. Therefore, in this paper we propose TransMUSE (Transferable Traffic Prediction in MUlti-Service Edge Networks), a novel deep learning framework that clusters similar services, groups edge-nodes into cohorts by traffic feature similarity, and employs a Transformer-based Multi-service Traffic Prediction Network (TMTPN), which can be directly transferred within a cohort without any customization. We demonstrate that TransMUSE exhibits imperceptible performance degradation in terms of mean absolute error (MAE) when forecasting traffic, compared with settings where a model is trained for each individual edge node. Moreover, our proposed TMTPN architecture outperforms the state-of-the-art, achieving up to 43.21% lower MAE in the multi-service traffic prediction task. To the best of our knowledge, this is the first work that jointly employs model transfer and multi-service traffic prediction to reduce measurement overhead, while providing fine-grained accurate demand forecasts for edge services provisioning.

High-speed wavelength interrogation system based on crossed HiBi-FLMs for array FBGs

A novel ultrafast wavelength interrogation system for array fiber Bragg gratings (FBGs) has been proposed and experimentally demonstrated, the operation principle is based on pulsed optical source (POS) combined with edge-filtering method and time-division multiplexing (TDM) technique. Thanks to the Gaussian-shape nature and numerous passband channels of the HiBi-FLM transmission, the simultaneous interrogation of array wavelength-division multiplexing (WDM) FBGs can be realized by employing crossed HiBi-FLMs scheme. The HiBi-FLM can provide an individual edge filter for each WDM-FBG and map the FBG reflection spectra into corresponding optical intensity, thus realizing the simultaneous interrogation of multiple WDM-FBGs. The interrogation speed of the proposed system is as high as 200 kHz, which can be further improved to hundreds of megahertz (MHz) by increasing the repetition rate of the POS, the response speed of the photodetector (PD), and the sampling rate of field programmable gate array (FPGA), as well as reducing the pulse width and the lengths of the delay lines. The proposed wavelength interrogation system features high speed, outstanding multiplexing capability, large dynamic range and simple configuration, thus providing a promising solution for various applications, such as quasi-distributed sensing, high-speed communication, medical treatment and the monitoring of ultrafast dynamic phenomena (e.g., molecular dynamics sensing and aircraft engine diagnostics).

Individual Tree Segmentation from Side-View LiDAR Point Clouds of Street Trees Using Shadow-Cut

Segmentation of vegetation LiDAR point clouds is an important method for obtaining individual tree structure parameters. The current individual tree segmentation methods are mainly for airborne LiDAR point clouds, which use elevation information to form a grid map for segmentation, or use canopy vertices as seed points for clustering. Side-view LiDAR (vehicle LiDAR and hand-held LiDAR) can acquire more information about the lower layer of trees, but it is a challenge to perform the individual tree segmentation because the structure of side-view LiDAR point clouds is more complex. This paper proposes an individual tree segmentation method called Shadow-cut to extract the contours of the street tree point cloud. Firstly, we separated the region of the trees using the binary classifier (e.g., support vector machine) based on point cloud geometric features. Then, the optimal projection of the 3D point clouds to the 2D image is calculated and the optimal projection is the case where the pixels of the individual tree image overlap the least. Finally, after using the image segmentation algorithm to extract the tree edges in the 2D image, the corresponding 3D individual tree point cloud contours are matched with the pixels of individual tree edges in the 2D image. We conducted experiments with the proposed method on LiDAR data of urban street trees, and the correctness, completeness, and quality of the proposed individual tree segmentation method reached 91.67%, 85.33%, and 79.19%, which were superior to the CHM-based method by 2.70%, 6.19%, and 7.12%, respectively. The results show that this method is a practical and effective solution for individual tree segmentation in the LiDAR point clouds of street trees.

Exposure theory for learning complex networks with random walks

AbstractRandom walks are a common model for the exploration and discovery of complex networks. While numerous algorithms have been proposed to map out an unknown network, a complementary question arises: in a known network, which nodes and edges are most likely to be discovered by a random walker in finite time? Here, we introduce exposure theory, a statistical mechanics framework that predicts the learning of nodes and edges across several types of networks, including weighted and temporal, and show that edge learning follows a universal trajectory. While the learning of individual nodes and edges is noisy, exposure theory produces a highly accurate prediction of aggregate exploration statistics.

Chiral cavity quantum electrodynamics

Cavity quantum electrodynamics, which explores the granularity of light by coupling a resonator to a nonlinear emitter1, has played a foundational role in the development of modern quantum information science and technology. In parallel, the field of condensed matter physics has been revolutionized by the discovery of underlying topological2,3,4, often arising from the breaking of time-reversal symmetry, as in the case of the quantum Hall effect. In this work, we explore the cavity quantum electrodynamics of a transmon qubit in a topologically nontrivial Harper–Hofstadter lattice5. We assemble the lattice of niobium superconducting resonators6 and break time-reversal symmetry by introducing ferrimagnets7 before coupling the system to a transmon qubit. We spectroscopically resolve the individual bulk and edge modes of the lattice, detect Rabi oscillations between the excited transmon and each mode and measure the synthetic-vacuum-induced Lamb shift of the transmon. Finally, we demonstrate the ability to employ the transmon to count individual photons8 within each mode of the topological band structure. This work opens the field of experimental chiral quantum optics9, enabling topological many-body physics with microwave photons 10,11 and providing a route to backscatter-resilient quantum communication.

Cross-kingdom co-occurrence networks in the plant microbiome: Importance and ecological interpretations.

Microbial co-occurrence network analysis is being widely used for data exploration in plant microbiome research. Still, challenges lie in how well these microbial networks represent natural microbial communities and how well we can interpret and extract eco-evolutionary insights from the networks. Although many technical solutions have been proposed, in this perspective, we touch on the grave problem of kingdom-level bias in network representation and interpretation. We underscore the eco-evolutionary significance of using cross-kingdom (bacterial-fungal) co-occurrence networks to increase the network’s representability of natural communities. To do so, we demonstrate how ecosystem-level interpretation of plant microbiome evolution changes with and without multi-kingdom analysis. Then, to overcome oversimplified interpretation of the networks stemming from the stereotypical dichotomy between bacteria and fungi, we recommend three avenues for ecological interpretation: (1) understanding dynamics and mechanisms of co-occurrence networks through generalized Lotka-Volterra and consumer-resource models, (2) finding alternative ecological explanations for individual negative and positive fungal-bacterial edges, and (3) connecting cross-kingdom networks to abiotic and biotic (host) environments.

How to improve the long-term quality of life, insomnia, and depression of survivors 10 years after the Wenchuan earthquake? A network analysis.

Previous studies have found a negative effect of depression and insomnia on the psychological health domain of quality of life (QOL) among earthquake survivors. However, little is known about the symptom-to-symptom interactions among the above psychological outcomes. This study thus aimed to assess the interplay among the above three variables in survivors 10 years after the Wenchuan earthquake at the symptom level. A total of 744 survivors completed the questionnaire at 10 years post-earthquake, reporting depressive symptoms, insomnia symptoms, and the psychological health domain of QOL. All network structures were estimated and compared using the network analysis approach in R version 4.1.1. Among the 744 survivors, 593 individuals did not have significant depressive and insomnia symptoms, while 151 individuals reported depressive and/or insomnia symptoms. "Little energy", "Suicidal ideation", and "Spirituality" were the key highest bridge symptoms in the three networks, respectively. Additionally, there were significant differences in network global strength, network structure, and individual edge weights between individuals with and without depression and/or insomnia. Intervention programs aimed at treating symptoms, such as exercise therapy, cognitive behavioural therapy, and spirituality education, may improve the QOL of survivors following an earthquake.

Network analysis of comorbid depression and anxiety and their associations with quality of life among clinicians in public hospitals during the late stage of the COVID-19 pandemic in China

BackgroundMental health problems are common among clinicians working in public hospitals even in the late stage of the COVID-19 pandemic. Network analysis is a novel approach to explore interactions between mental health problems at the symptom level. This study examined the network structure of comorbid depression and anxiety and their associations with quality of life (QOL) among hospital clinicians in China during the late stage of the COVID-19 pandemic. MethodsA total of 4931 participants were recruited from October 13 to 22, 2020. The nine-item Patient Health Questionnaire (PHQ-9), seven-item Generalized Anxiety Disorder Scale (GAD-7), and the World Health Organization Quality of Life Questionnaire-Brief Version (WHOQOL-BREF) were used to measure depressive and anxiety symptoms, and QOL, respectively. Central and bridge symptoms were identified with centrality and bridge centrality indices, respectively. Network stability was examined using the case-dropping procedure. ResultsThe prevalence of depression (defined as PHQ-9 total score ≥ 5) was 35.1 % [95 % confidence interval (CI) = 33.73–36.41 %)], the prevalence of anxiety (GAD-7 total score ≥ 5) was 32.5 % (95 % CI = 31.20–33.84 %), while the prevalence of comorbid depression and anxiety was 26.9 % (95 % CI = 25.7–28.2 %). “Impaired motor skills”, “Trouble relaxing” and “Uncontrollable worry” were the central symptoms in the whole depression-anxiety network. “Irritability”, “Feeling afraid” and “Sad mood” were the most key bridge symptoms linking depression and anxiety. Three symptoms (“Fatigue”, “Trouble relaxing” and “Nervousness”) were the most strongly and negatively associated with QOL. Neither gender nor the experiences of caring for COVID-19 patients was associated with network global strength, distribution of edge weights or individual edge weights. LimitationsThe causality between variables could not be established. Depressive and anxiety symptoms were assessed by self-report measures, which may result in recall bias and limitations in capturing clinical phenomena. ConclusionsBoth the central (i.e., “Impaired motor skills”, “Trouble relaxing” and “Uncontrollable worry”) and bridge symptoms (i.e., “Irritability”, “Feeling afraid” and “Sad mood”) identified in this network analysis should be targeted in specific treatment and preventive measures for comorbid depressive and anxiety symptoms among clinicians in the late stage of the pandemic. Furthermore, “Fatigue”, “Trouble relaxing” and “Nervousness” are key symptoms to address to improve clinicians' QOL.

From Edge City to City?

Problem, research strategy and findings This century’s planning challenge is one of retrofitting suburbia. As relatively compact suburban forms, edge cities may be the most amenable to retrofit. In the 30 years since the publication of Garreau’s Edge City, to what extent and in what ways have planning intentions changed to favor greater urbanity in edge cities? Across 10 indicators, we examined planning intentions covering 117 of 123 edge cities identified in 1991. Content analysis of plans revealed the mixed intentions for individual edge cities and for different edge cities across metropolitan regions, but also strong and consistent intentions for a large minority of edge cities we label cities in the making. Our findings underline the limited appeal of the concept to the public, politicians, and planners. Takeaways for practice Across a range of indicative intentions, planning frameworks can be strengthened with a view to fostering greater urbanity in edge cities. Plans covering edge cities continue to lack the spatial and numeric specificity that would help ensure better implementation of stated intentions. Plans covering edge cities could usefully have a more joined-up or consistent approach across different but interrelated intentions.

Multi-active Site Dynamics on a Molecular Cr/Co/Se Cluster Catalyst.

This study uncovers the interconnected reactivity of the three catalytically active sites of an atomically precise nanocluster Cr3(py)3Co6Se8L6 (1(py)3, L = Ph2PNTol-, Ph = phenyl, Tol = 4-tolyl). Catalytic and stoichiometric studies into tosyl azide activation and carbodiimide formation enabled the isolation and crystallographic characterization of key catalytically competent metal-imido intermediates, including the tris(imido) cluster 1(NTs)3, the catalytic resting state 1(NTs)3(CNtBu)3, and the site-differentiated mono(imido) cluster 1(NTs)(CNtBu)2. In the stoichiometric regime, nitrene transfer proceeds via a stepwise mechanism, with the three active sites engaging sequentially to produce carbodiimide. Moreover, the chemical state of neighboring active sites was found to regulate the affinity for substrates of an individual Cr-imido edge site, as revealed by comparative structural analysis and CNtBu binding studies.

Resistivity of Surface Steps in Bulk-Insulating Topological Insulators

Electron transport in topological insulators usually involves both topologically protected surface states and trivial electronic states in the bulk material. The surface transport is particularly interesting; however, it is also susceptible to atomic defects on the surfaces, such as vacancies, impurities, and step edges. Experimental determination of scattering effects of these surface defects requires both nanoscale spatial resolution and the ability to decipher surface transport from bulk transport. Here we directly measure the resistivity of individual surface steps in the surface dominating transport process of topological insulator Bi2Te2Se. A variable probe-spacing transport spectroscopy with a multiprobe scanning tunneling microscope is used to differentiate the surface conductance from bulk conductance, allowing the identification of a surface dominant transport regime. The technique also reveals a deviation from ideal 2D transport at atomic steps. Then, a multi-probe scanning tunneling potentiometry is employed to visualize the electrochemical potentials across individual step edges. A quantitative analysis of the potential distributions enables us to acquire a resistivity of 0.530 mΩ · cm for the one quintuple-layer atomic step. The result indicates that atomic defects, despite preserving the time-reversal symmetry, can still significantly affect the transport in topological insulators.

Transmission Power Control for Over-the-Air Federated Averaging at Network Edge

<i>Over-the-air computation</i> (AirComp) has emerged as a new analog power-domain <i>non-orthogonal multiple access</i> (NOMA) technique for low-latency model/gradient-updates <i>aggregation in federated edge learning</i> (FEEL). By integrating communication and computation into a joint design, AirComp can significantly enhance the communication efficiency, but at the cost of aggregation errors caused by channel fading and noise. This paper studies a particular type of FEEL with federated averaging (FedAvg) and AirComp-based model-update aggregation, namely <i>over-the-air</i> FedAvg (Air-FedAvg). We investigate the transmission power control in Air-FedAvg to combat against the AirComp aggregation errors for enhancing the training accuracy and accelerating the training speed. Towards this end, we first analyze the convergence behavior (in terms of the optimality gap) of Air-FedAvg with aggregation errors at different outer iterations. Then, to enhance the training accuracy, we minimize the optimality gap by jointly optimizing the transmission power control at edge devices and the denoising factors at edge server, subject to a series of power constraints at individual edge devices. Furthermore, to accelerate the training speed, we also minimize the training latency of Air-FedAvg with a given targeted optimality gap, in which learning hyper-parameters including the numbers of outer iterations and local training epochs are optimized jointly with the power control. Finally, numerical results show that the proposed transmission power control policy achieves significantly faster convergence speed for Air-FedAvg, as compared with benchmark policies with fixed power transmission or per-iteration <i>mean squared error</i> (MSE) minimization. It is also shown that the Air-FedAvg achieves an order-of-magnitude shorter training latency than the conventional FedAvg with digital <i>orthogonal multiple access</i> (OMA-FedAvg).

Wavelength-tunable dual-band edge plasmon mode based on gold edge-hole plasmonic nanostructure

Metal nanostructures have attracted attention because of their unique optical characteristics derived from localized surface plasmon resonance. In this study, gold nanoholes with edge structures were designed and characterized to possess two plasmon bands at different wavelengths. These bands were assigned to the plasmon modes that generated a concentrated electric field at the tip of the individual edge structures. Furthermore, wavelengths of these bands were independently tuned from the visible to near-infrared wavelengths with shape and size of the gold nanoholes and edge structures as variables. To demonstrate wavelength tuning, a gold nanostructure with edge plasmon bands at wavelengths of approximately 785 and 1064 nm (typical laser wavelengths for surface-enhanced Raman scattering and optical tweezers, respectively) was successfully obtained via optical simulation. The results indicate that the structure designed in this study provides a guideline for the design of optical devices with efficient multiple optical functions.

An Exploration of Self-Reported Sleep Inertia Symptoms Using Network Analysis.

PurposeSleep inertia (SI) is the transitional state accompanied by compromised cognitive and physical performance and sleepiness. Network analysis offers a potential new framework to conceptualize a complex network of symptom–symptom interactions, and the network structure is analyzed to reveal the core characteristics. However, no previous study examined the network structure of SI symptoms. Thus, this study aimed to elucidate characteristics and compare sex differences of SI symptom networks in the general population.Materials and MethodsA total of 1491 participants from China were recruited from 30 May to 17 June, 2021. SI symptoms were assessed by using the Sleep Inertia Questionnaire (SIQ). The network structures were estimated and compared using network analytic methods in the R version 4.1.1.ResultsCentrality properties analysis of the expected influence suggested that symptoms of “Feel sleepy”, “Groggy, fuzzy or hazy mind”, and “Dread starting your day” exerted greatest influences. The weighted adjacency matrix revealed that the “Dread starting your day” and “Anxious about the upcoming day” edge showed the strongest connection (edge weight value = 0.70). The network comparison test found no significant difference in network global strength (p=0.928), distribution of edge weights (p=0.194) and individual edge weights (all p values >0.05 after Holm–Bonferroni corrections) between males and females.ConclusionSymptoms of “Feel sleepy”, “Groggy, fuzzy or hazy mind”, and “Dread starting your day” were central in the SI symptom network. Intervention, such as the artificial dawn and change in body temperature, for symptoms of “Feel sleepy”, “Groggy, fuzzy or hazy mind”, and “Dread starting your day” might be crucial to hasten the dissipation of SI in the general population who may need to perform tasks upon waking.

Mapping Network Connectivity Among Symptoms of Depression and Pain in Wuhan Residents During the Late-Stage of the COVID-19 Pandemic.

BackgroundSymptoms of depression and pain often overlap, and they negatively influence the prognosis and treatment outcome of both conditions. However, the comorbidity of depression and pain has not been examined using network analysis, especially in the context of a pandemic. Thus, we mapped out the network connectivity among the symptoms of depression and pain in Wuhan residents in China during the late stage of the COVID-19 pandemic.MethodsThis cross-sectional study was conducted from May 25, 2020 to June 18, 2020 in Wuhan, China. Participants' depressive and pain symptoms were assessed using the 9-item Patient Health Questionnaire (PHQ9) and a pain numeric rating scale (NRS), respectively. Network analyses were performed.ResultsIn total, 2,598 participants completed all assessments. PHQ4 (fatigue) in the depression community showed the highest strength value, followed by PHQ6 (worthlessness) and PHQ2 (depressed or sad mood). PHQ4 (fatigue) was also the most key bridge symptom liking depression and pain, followed by PHQ3 (sleep difficulties). There were no significant differences in network global strength (females: 4.36 vs. males: 4.29; S = 0.075, P = 0.427), network structure-distribution of edge weights (M = 0.12, P = 0.541), and individual edge weights between male and female participants.ConclusionDepressive and pain symptoms showed strong cross-association with each other. “Fatigue” was the strongest central and bridge symptom in the network model, while “sleep difficulties” was the second strongest bridge symptom. Targeting treatment of both fatigue and sleep problems may help improve depressive and pain symptoms in those affected.

Edge-based local push for personalized PageRank

Personalized PageRank (PPR) is a popular node proximity metric in graph mining and network research. A single-source PPR (SSPPR) query asks for the PPR value of each node on the graph. Due to its importance and wide applications, decades of efforts have been devoted to the efficient processing of SSPPR queries. Among existing algorithms, LocalPush is a fundamental method for SSPPR queries and serves as a cornerstone for subsequent algorithms. In LocalPush , a push operation is a crucial primitive operation, which distributes the probability at a node u to ALL u 's neighbors via the corresponding edges. Although this push operation works well on unweighted graphs, unfortunately, it can be rather inefficient on weighted graphs. In particular, on unbalanced weighted graphs where only a few of these edges take the majority of the total weight among them, the push operation would have to distribute "insignificant" probabilities along those edges which just take the minor weights, resulting in expensive overhead. To resolve this issue, in this paper, we propose the EdgePush algorithm, a novel method for computing SSPPR queries on weighted graphs. EdgePush decomposes the aforementioned push operations in edge-based push , allowing the algorithm to operate at the edge level granularity. As a result, it can flexibly distribute the probabilities according to edge weights. Furthermore, our EdgePush allows a fine-grained termination threshold for each individual edge, leading to a superior complexity over LocalPush. Notably, we prove that EdgePush improves the theoretical query cost of LocalPush by an order of up to O ( n ) when the graph's weights are unbalanced. Our experimental results demonstrate that EdgePush significantly outperforms state-of-the-art baselines in terms of query efficiency on large motif-based and real-world weighted graphs.

A review on federated learning towards image processing

Nowadays, data privacy is an important consideration in machine learning. This paper provides an overview of how Federated Learning can be used to improve data security and privacy. Federated Learning is made up of three distinct architectures that ensure that privacy is never jeopardised. Federated learning is a type of collective learning in which individual edge devices are trained and then aggregated on the server without sharing edge device data. On the other hand, federated learning provides secure models with no data sharing, resulting in a highly efficient privacy-preserving solution that also provides security and data access. We discuss the various frameworks used in federated learning, as well as how federated learning is used with machine learning, deep learning, and datamining. This paper focuses on image processing applications that ensure that data trained on the model is secure and protected. We provide a comprehensive overview of the key issues raised in recent literature, as well as an accurate description of the related research work.

A comparative study of biopolymer adsorption on model anisotropic clay surfaces using quartz crystal microbalance with dissipation (QCM-D)

Mitigation of colloid clay particles is critical during flotation and flocculation processes in mineral processing. Most organic and inorganic mitigation reagents have negative impacts on the environment and human health; therefore, biologically derived substances have been attracting attention as alternative reagents. Given the anisotropic nature of clay surfaces, it is imperative to understand reagent adsorption on the individual edge and basal plane surfaces of clays. Quartz crystal microbalance with dissipation (QCM-D) was used in this study to determine the adsorption characteristics of three biopolymers (the protein-based biopolymer, lysozyme, and protein and polysaccharide-based oligomers; protamine and pectin) on model surfaces of the anisotropic edge and basal planes of the clays (e.g., kaolinite and serpentine). SiO2 sensor representing the tetrahedral basal plane, Al2O3 and Mg(OH)2 representing the octahedral basal planes, AlSiO and MgSiO representing the edge faces of clays were used as model surfaces of clay minerals. For kaolinite, protamine adsorbed preferentially on the silica (SiO2) tetrahedral surface at pH 7 and on the alumina (Al2O3) surface at pH 10. Protamine adsorbed primarily on magnesium hydroxide (Mg(OH)2), representative of serpentine, at pH 7 and 10. Lysozyme adsorbed preferentially and irreversibly on the edge basal plane surfaces of both clays at pH 10, while it showed a higher affinity for octahedral surfaces (alumina and magnesium hydroxide) at pH 7. In contrast, pectin adsorbed strongly on the magnesium hydroxide, representative of the basal plane surface of serpentine. An adsorption study revealed that electrostatic attraction and/or hydrogen bonding mechanisms contributed to the adsorption of biopolymers on clay surfaces. This investigation provides a fundamental and practical understanding of biopolymer interactions with clay surfaces during selective flotation and flocculation.

Inferring the Genetic Influences on Psychological Traits Using MRI Connectivity Predictive Models: Demonstration with Cognition

Introduction: Genetic correlations between brain and behavioral phenotypes in analyses from major genetic consortia have been weak and mostly nonsignificant. fMRI models of systems-level brain patterns may help improve our ability to link genes, brains, and behavior by identifying reliable and reproducible endophenotypes. Work using connectivity-based predictive modeling has generated brain-based proxies of behavioral and neuropsychological variables. If such models capture activity in inherited brain systems, they may offer a more powerful link between genes and behavior. Method: As a proof of concept, we develop models predicting intelligence (IQ) based on fMRI connectivity and test their effectiveness as endophenotypes. We link brain and IQ in a model development dataset of N = 3,000 individuals and test the genetic correlations between brain models and measured IQ in a genetic validation sample of N = 13,092 individuals from the UK Biobank. We compare an additive connectivity-based model to multivariate LASSO and ridge models phenotypically and genetically. We also compare these approaches to single “candidate” brain areas. Results: We found that predictive brain models were significantly phenotypically correlated with IQ and showed much stronger correlations than individual edges. Further, brain models were more heritable (h2 = 0.155–0.181) than single brain regions (h2 = 0.038–0.118) and captured about half of the genetic variance in IQ (rG = 0.422–0.576), while rGs with single brain measures were smaller and nonsignificant. For the different approaches, LASSO and ridge were similarly predictive, with slightly weaker performance of the additive model. LASSO model weights were highly theoretically interpretable and replicated known brain IQ associations. Finally, functional connectivity models trained in midlife showed genetic correlations with early life correlates of IQ, suggesting some stability in the prediction of fMRI models. Conclusion: Multisystem predictive models hold promise as imaging endophenotypes that offer complex and theoretically relevant conclusions for future imaging genetics research.