Ideal Network Research Articles

Unsupervised graph denoising via feature-driven matrix factorization

Graph denoising is becoming a promising solution for robust graph embedding, which aims to construct an ideal network by removing noisy edges. Currently, the mainstream denoising approaches often build upon the low-rank assumption by removing high-rank harmful edges with singular value decomposition. Although this approach often allows yielding good graph embedding, the high time complexity limits its application to large-scale networks. In this paper, we propose an effective and efficient algorithm for robust graph embedding: Latent Feature-driven Graph Denoising (LFGD). The basic idea is to leverage node latent features to construct an ideal low-rank network by exploiting the relationship between topology and feature information. To this end, the original features are first mapped into a latent space, and then a low-rank network is reconstructed by imposing the semantic preservation loss and structure loss (including regression loss and sparsity constraint). In addition, we propose a sampling-based strategy to further speed up the proposed method, which finally results in linear time complexity. LFGD works independently of downstream task, which makes the denoised structure more general and reliable. Extensive experiments have demonstrated the superiority of the LFGD to many state-of-the-art algorithms under various attacks in terms of node classification, robustness and running time. Our code is available at: https://github.com/hanwangme/LFGD.

Crosslinker energy landscape effects on dynamic mechanical properties of ideal polymer hydrogels.

Reversible crosslinkers can enable several desirable mechanical properties, such as improved toughness and self-healing, when incorporated in polymer networks for bioengineering and structural applications. In this work, we performed coarse-grained molecular dynamics to investigate the effect of the energy landscape of reversible crosslinkers on the dynamic mechanical properties of crosslinked polymer network hydrogels. We report that, for an ideal network, the energy potential of the crosslinker interaction drives the viscosity of the network, where a stronger potential results in a higher viscosity. Additional topographical analyses reveal a mechanistic understanding of the structural rearrangement of the network as it deforms and indicate that as the number of defects increases in the network, the viscosity of the network increases. As an important validation for the relationship between the energy landscape of a crosslinker chemistry and the resulting dynamic mechanical properties of a crosslinked ideal network hydrogel, this work enhances our understanding of deformation mechanisms in polymer networks that cannot easily be revealed by experiment and reveals design ideas that can lead to better performance of the polymer network at the macroscale.

Effect of elemental substitution on transition threshold behaviours of Ge-As(Sb)-Se glasses

In this paper, We prepare two groups of glasses: one is Ge<sub><i>x</i></sub>As<sub>20</sub>Se<sub>80–<i>x</i></sub> with x ranging from 5% to 32.5%, the other is Ge<sub><i>x</i></sub>Sb<sub>20</sub>Se<sub>80–<i>x</i></sub> with x spanning from 5% to 25%, by using the conventional melt-quench method, and investigate the effect of the elemental substitution of Sb for As on the threshold behaviors in Ge<sub><i>x</i></sub>As(Sb)<sub>20</sub>Se<sub>80–<i>x</i></sub> glasses. We are to understand to what extent the topological model and chemical order model can explain the correlation between physical properties and glass compositions, and how the chemical composition can affect the glass transition threshold. Glass transition temperature is measured by the differential scanning calorimeter (Mettler-Toledo, DSC1) with different scanning rates ranging from 5 K/min to 30 K/min under a uniform nitrogen gas flow of 50 mL/min, the glass density is measured by a Mettler H20 balance with a MgO crystal used as a reference. Samples of each glass composition are weighed five times and the average density is recorded. The refractive index of the glass at 1.5 um is measured by a Metricon Model 2010 prism coupler. Raman spectra are measured by a T64000 Jobin-Yvon-Horiba micro-Raman spectrometer equipped with a liquid-nitrogen-cooled CCD detector. The 830 nm laser line is used as an excitation source, and the laser power is kept as small as possible to avoid any photo-induced effects. The resolution of the spectrometer is about 0.5 cm<sup>–1</sup>. The systematic measurements of these physical parameters show that while the transition thresholds at MCN = 2.4 and 2.67 are verified in the Ge-As-Se glasses with ideal covalent network, these two transitions represent the covalent network structure inside the glass from an under-constrained “floppy” network to an over-constrained “rigid” phase and from the two-dimensional to the three-dimensional “stressed rigid” phase respectively. However, when As is substituted by Sb, the the resulting Ge<sub><i>x</i></sub>Sb<sub>20</sub>Se<sub>80–<i>x</i></sub> glass with non-ideal covalent network will change its transition threshold, changing into the chemically stoichiometric composition. We further deconvolve Raman scattering spectra into different structural units and the change of their respective intensity shows the same behavior, which is ascribed to the chemical effect induced by a large difference in atomic radius between As and Sb, and a relatively strong ionic feature of element Sb.

Pathways to school improvement: Discovering network patterns of school principals

Abstract The purpose of this study is to examine the network effect of school principals as it relates to school improvement. Network practices of school principals are compared to an innovative practice for improving networking practices. Through descriptive statistics and chi-square goodness of fit, we illustrate the difference between what school principals do concerning their networking practices for school improvement compared to an innovative ideal approach for using network working for school improvement. Findings indicate there is a statistically significant difference between school principals’ networking practices in comparison to ideal networking practices for school improvement. There are also differences between who school principals seek out for ideas and who they seek out for feedback concerning their school improvements. Further discussion informs how the next generation of school principals can be equipped with innovative skills for tackling 21st-century school improvement issues.

Federated and Online Dynamic Spectrum Access for Mobile Secondary Users

Users in dynamic spectrum access (DSA) with federated reinforcement learning (FRL) autonomously access channels, avoiding centralized coordination and protecting users’ privacy. However, existing FRL-based DSA mechanisms are limited to ideal network states, i.e., assuming that channel states and users’ interference relationships are unchanged. Besides, users should upload intermediate results simultaneously for federated aggregation. The above conditions are impractical for mobile users since their network states and locations are unstable. Meanwhile, newly connected users have to train their models through local data with numerous computing resources since global models are unsuitable for them. We propose FRDSA, an FRL-based secure and lightweight channel selection mechanism in DSA for mobile users under dynamic network states. An independent channel selection environment with a virtual group strategy is presented to avoid interference between users under unstable channel states. Furthermore, an asynchronous parameter aggregation method in FRDSA dynamically adjusts the aggregation factors without users simultaneously uploading intermediate results. Simulations based on real trajectory data show that FRDSA significantly reduces approximately 60% interference between mobile users under unstable network states. Newly connected users can directly apply the well-trained global model to access channels autonomously instead of retraining a model, effectively reducing mobile users’ computing resource requirements.

Impact of Data-Augmentation on Brain Tumor Detection Using Different YOLO Versions Models

Impact of Data-Augmentation on Brain Tumor Detection Using Different YOLO Versions Models

Cooperative optimal operation of hybrid energy integrated system considering multi‐objective dragonfly algorithm

AbstractTo meet the recent energy demand, hybrid energy systems (HESs) play an important role in providing more stable power as well as backup power to the grid. It is always preferable for the power system network to be operated in a secure, dependable, stable, and cost‐effective manner. The optimal power flow (OPF) problem is used to determine the ideal power system network control parameter settings. The majority of the OPF problem, fuel cost reduction is treated as an objective function, but the environmental impact of the generating is ignored. Both fuel expense and emission are treated as objective functions of the work in this case. This research describes a solution for multi‐objective optimal power flow (MOOPF) with a HES integrated conventional power system. HES in this work is made up of a wind park and a compressed air energy storage system (CAES). In this case, locational marginal pricing (LMP) is employed to determine the best location of HES in the power system. To achieve the goals of this work, the multi‐objective dragonfly algorithm (MODA) is used. The IEEE 30 bus system is used to evaluate the MODA. The MODA method findings are validated against other well‐known optimization algorithms such as the multi‐objective ant colony system (MOACS) algorithm, multi‐objective enhanced self‐adaptive differential evolution (MOESDE) algorithm, modified multi‐objective evolutionary algorithm‐based decomposition (MOEA/D), multi‐objective particle swarm algorithm (MOPOSO), and non‐dominated sorting genetic algorithm (NSGA‐II).

Regularized deep learning for unsupervised random noise attenuation in poststack seismic data

Abstract Deep learning methods achieve excellent noise reduction performances in seismic data processing compared with traditional methods. However, deep learning usually requires a large number of pairwise noisy-clean training data, which is an extremely challenging task. In this paper, an unsupervised approach without clean seismic data is proposed to suppress random noise. Seismic data is divided into odd and even traces, which serve as the input and output of the depth network, so that the proposed algorithm can be trained directly on the original data. What is more, the proposed method introduces two regularization terms to solve the over-smoothing problem caused by reconstruction of adjacent traces. The first term considers an ideal denoising network that does not cause oversmooth as a constraint, while the second term considers the structural information existing in seismic data. Experiments on synthetic post-stack data illustrate that the proposed method obtain a higher signal-to-noise ratio than the comparison methods. In the application of field post-stack seismic data, the proposed method can effectively maintain the seismic amplitude and generate good spectral characteristics.

Deep Reinforcement Learning Framework with Q Learning For Optimal Scheduling in Cloud Computing

Cloud computing is an emerging technology that is increasingly being appreciated for its diverse uses, encompassing data processing, The Internet of Things (IoT) and the storing of data. The continuous growth in the number of cloud users and the widespread use of IoT devices have resulted in a significant increase in the volume of data being generated by these users and the integration of IoT devices with cloud platforms. The process of managing data stored in the cloud has become more challenging to complete. There are numerous significant challenges that must be overcome in the process of migrating all data to cloud-hosted data centers. High bandwidth consumption, longer wait times, greater costs, and greater energy consumption are only some of the difficulties that must be overcome. Cloud computing, as a result, is able to allot resources in line with the specific actions made by users, which is a result of the conclusion that was mentioned earlier. This phenomenon can be attributed to the provision of a superior Quality of Service (QoS) to clients or users, with an optimal response time. Additionally, adherence to the established Service Level Agreement further contributes to this outcome. Due to this circumstance, it is of utmost need to effectively use the computational resources at hand, hence requiring the formulation of an optimal approach for task scheduling. The goal of this proposed study is to find ways to allocate and schedule cloud-based virtual machines (VMs) and tasks in such a way as to reduce completion times and associated costs. This study presents a new method of scheduling that makes use of Q-Learning to optimize the utilization of resources.The algorithm's primary goals include optimizing its objective function, building the ideal network, and utilizing experience replay techniques.

Investigating Pathways to Minimize Sensor Power Usage for the Internet of Remote Things.

The Internet of Remote Things (IoRT) offers an exciting landscape for the development and deployment of remote wireless sensing nodes (WSNs) which can gather useful environmental data. Low Power Wide Area Networks (LPWANs) provide an ideal network topology for enabling the IoRT, but due to the remote location of these WSNs, the power and energy requirements for such systems must be accurately determined before deployment, as devices will be running on limited energy resources, such as long-life batteries or energy harvesting. Various sensor modules that are available on the consumer market are suitable for these applications; however, the exact power requirements and characteristics of the sensor are often not stated in datasheets, nor verified experimentally. This study details an experimental procedure where the energy requirements are measured for various sensor modules that are available for Arduino and other microcontroller units (MCUs). First, the static power consumption of continually powered sensors was measured. The impact of sensor warm-up time, associated with powering on the sensor and waiting for reliable measurements, is also explored. Finally, the opportunity to reduce power for sensors which have multiple outputs was investigated to see if there is any significant reduction in power consumption when obtaining readings from fewer outputs than all that are available. It was found that, generally, CO2 and soil moisture sensors have a large power requirement when compared with temperature, humidity and pressure sensors. Limiting multiple sensor outputs was shown not to reduce power consumption. The warm-up time for analog sensors and digital sensors was generally negligible and in the order of 10-50 ms. However, one CO2 sensor had a large overhead warm-up time of several seconds which added a significant energy burden. It was found that more, or as much, power could be consumed during warm-up as during the actual measurement phase. Finally, this study found disparity between power consumption values in datasheets and experimental measurements, which could have significant consequences in terms of battery life in the field.

Influence of Liquid Splitting Behavior at Intersections on Infiltration Dynamics in an Unsaturated Fracture Network

AbstractUnderstanding the macro‐scale flow characteristics in the fractured vadose zone is of great importance for subsurface hydrological and environmental applications. Here we develop an idealized fracture network model composed of a series of linked intersections, aiming to reveal the roles of local fluid flow, storage and splitting behaviors at intersections in controlling macroscopic unsaturated flow in a fracture network. By setting different local flow rules, unsaturated flow dynamics and structure in networks are systematically investigated. Numerical simulations suggest that avalanche infiltration mode, that is, sudden release of a large amount of water from the network, emerges spontaneously, and is modulated by the local splitting behavior. In terms of water infiltration structure, we find that uneven liquid splitting at intersections inevitably leads to preferential flow in networks. The preferential paths strongly rely on the competition between gravitational and capillary forces, relating to network structure and water saturation. When flow is dominated by gravity, preferential paths extend along large‐aperture routes; conversely, when flow is dominated by capillarity, the small‐aperture paths are preferred. In terms of infiltration dynamics, we show that the power spectral density of the water saturation time series in the network follows a power law with an exponent of −2 for all simulations with different structural parameters and local flow rules, suggesting a universal self‐organized criticality behavior for unsaturated flow in fractured rocks. The improved understanding from this study may shed new light on the complex flow dynamics for unsaturated flow in fractured media.

Coexistence of Crumpling and Flat Sheet Conformations in Two-Dimensional Polymer Networks: An Understanding of Aggrecan Self-Assembly.

We investigate the conformational properties of self-avoiding two-dimensional (2D) ideal polymer networks with tunable mesh sizes as a model of self-assembled structures formed by aggrecan. Polymer networks having few branching points and large enough mesh tend to crumple, resulting in a fractal dimension of d_{f}≈2.7. The flat sheet behavior (d_{f}=2) emerges in 2D polymer networks having more branching points at large length scales; however, it coexists with crumpling conformations at intermediate length scales, a feature found in scattering profiles of aggrecan solutions. Our findings bridge the long-standing gap between theories and simulations of polymer sheets.

Automated Brain Tumor Detection using Ideal Shallow Neural Network with Artificial Jellyfish Optimization.

Brain tumors are predicted from Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) scan images. In recent years, image processing-based automated tools are developed to predict tumor areas with less human interference. However, such automated tools are suffering from computational complexity and reduced accuracy in certain critical images. In the proposed work, an Ideal Shallow Neural Network (ISNN) is utilized to improve the prediction accuracy, and the computational complexity is reduced by implementing an Artificial Jellyfish Optimization (AJO) algorithm for minimizing the feature dimensionality. The proposed method utilizes MRI images for the verification process as they are more informative than the CT scan image. The BRATS and the Kaggle datasets are used in this work and a Gabor filtering technique is used for noise reduction and a histogram equalization is used for enhancing the tumor boundary regions. The classification results observed from the AJO-ISNN are further forwarded towards the segmentation process and which uses the Centroid Weighted Segmentation (WCS) along with a Grasshopper Optimization Algorithm (GOA) for improving the segmentation over the boundary regions of the brain tumor. The experimental result indicates a classification accuracy of 95.14% on the proposed AJO-ISNN model and AJO-ISNN is comparatively better than the Convolutional Neural Network (CNN) model accuracy of 85.41% and VGG 19 model accuracy of 93.75% while implemented with the AJO optimization model. Similarly, the Dice Similarity Coefficient of the proposed CWS-GOA also reaches 93.15% when performed with both BRATS and Kaggle datasets. Apart from the accuracy attainments the proposed work classifies and segments the tumor region in around 65 seconds on average of 200 image verifications and that is comparatively better than the previous multi-cascaded CNN and the InceptionV3 models.

Hybrid Chameleon Search and Remora Optimization Algorithm‐based Dynamic Heterogeneous load balancing clustering protocol for extending the lifetime of wireless sensor networks

SummaryClustering is an indispensable strategy that helps towards the extension of lifetime of each sensor nodes with energy stability in wireless sensor networks (WSNs). This clustering process aids in sustaining energy efficiency and extended network lifetime in sensitive and critical real‐life applications that include landslide monitoring and military applications. The dynamic characteristics of WSNs and several cluster configurations introduce challenge in the process of searching an ideal network structure, a herculean challenge. In this paper, Hybrid Chameleon Search and Remora Optimization Algorithm‐based Dynamic Clustering Method (HCSROA) is proposed for dynamic optimization of wireless sensor node clusters. It utilized the global searching process of Chameleon Search Algorithm for selecting potential cluster head (CH) selection with balanced trade‐off between intensification and extensification. It determines an ideal dynamic network structure based on factors that include quantity of nodes in the neighborhood, distance to sink, predictable energy utilization rate, and residual energy into account during the formulation of fitness function. It specifically achieved sink node mobility through the integration of the local searching capability of Improved Remora Optimization Algorithm for determining the optimal points of deployment over which the packets can be forwarded from the CH of the cluster to the sink node. This proposed HCSROA scheme compared in contrast to standard methods is identified to greatly prolong network lifetime by 29.21% and maintain energy stability by 25.64% in contrast to baseline protocols taken for investigation.

A model with contact maps at both polymer chain and network scales for tough hydrogels with chain entanglement, hidden length and unconventional network topology

Tough hydrogels with unconventional polymer network architectures often show superior mechanical properties because of the microstructures at the polymer chain and network scales. In this study, a single-chain based model, or called the network topology with hidden length and entanglement (NeTHE) model for the tough hydrogels with chain entanglement, hidden length and unconventional network topology, is developed by imposing contact maps at both polymer chain and network scales on the Arruda-Boyce model. In particular, the contact map is imposed at the polymer chain scale to characterize both entanglement and hidden length simultaneously, while the contact map is imposed at the polymer network scale to consider different polymeric network topologies. The present NeTHE model is validated well by comparing published experimental results of a highly entangled hydrogel and two soft regular hydrogels with hidden lengths. It is confirmed that the present model is able to characterize the hardening, softening, swelling, and damage of the hydrogels. By simplifications, the NeTHE model is reduced directly to the pseudo-elasticity model [1]. Furthermore, the model is able to cover the exponential or scaling law type of damage behaviors of the hydrogels. Finally, several parameter studies are conducted for the swelling and hysteresis of the hydrogels. It is observed theoretically that the constraints of polymer chains due to entanglements result in the high swelling resistance of highly entangled hydrogels, while the hidden length facilitates the swelling of soft regular hydrogels. It is also found that, regarding the hysteresis of hydrogels with different polymer network topologies, the ideal polymer network possesses the highest damage tolerance when compared with both the random and high-functionality crosslinked polymer networks. Regarding the effects of entanglements and hidden lengths on the hysteresis, the highly entangled hydrogel shows a lower stress–stretch hysteresis due to the energy dissipated partially to harden polymer chains when compared with the soft regular hydrogel of which the energy is dissipated fully to fracture the polymer chains.

Re-conceptualizing the ideal homes in rural China: an actor-network theory approach

With the shift in relational ontology, the concept of home is interpreted as a dynamic and relational process, in which both human and non-human elements need to be valued. Adopting ANT (actor-network theory) as an approach, this article reveals the interactive relationship and translation process between urban-rural migrants and rural heterogeneous actors (including farmland, landscapes, local people, means of subsistence, and technological facilities) in the connection of ideal home networks. It involves three steps of translation between actors, namely problem presentation of key actors, enrollment and benefit granting by key actors, and negotiation of “obligatory passage points” (OPP) based on actors’ common goals. Based on a two-year qualitative research of 18 interviewees, this article demonstrates how migrants adapt to the countryside through enrollment and benefit granting, and how sustainability and recyclability become common interests and OPP for actors in homemaking. The article argues that ANT provides a descriptive focus on how rural life has transformed from a backward neighborhood or abstract nostalgia to a harmonious new home that blends traditional culture, ecological ideas, and modern technology in the context of China. The study also points out that in the process of translating interests between actors, the human-land relationship still shows complexity and instability. Future research needs to pay more attention to the role of government officials and governance policies in the building of home networks.

Study of steel buildings with LCF system under critical mainshock-aftershock sequence: Evaluation of fragility curves and estimation of the response modification factor by artificial intelligence

In the seismic active zones, the stored energy in faults is continuously released which can lead to the seismic sequence phenomenon. So that, strong main shocks may involve numerous foreshocks and aftershocks with a magnitude close to the main shock. Hence, it is expected that the behavior of buildings under successive shocks would differ compared to one earthquake. Thus, this paper examined the effects of the seismic sequence phenomenon on the performance of steel buildings with the Linked Column Frame (LCF) system under critical successive earthquakes. As LCF is relatively new lateral force resisting system and considering the successive scenarios still seems necessary, this paper estimates the Response modification factors (R factors) by artificial intelligence, besides evaluating the fragility curves under critical mainshock-aftershock sequence. Therefore, building frames with 3, 5, 7, 9, and 11 stories equipped with the LCF lateral load resisting system had been analyzed in the OpenSees software consist of incremental dynamic, nonlinear static, and linear and nonlinear dynamic analyses. R factors were calculated under the seismic scenarios with/without sequence after performing more than 18 thousand nonlinear dynamic analyses. For more comprehensive examination of R factors, in addition to the number of stories, the effect of other parameters, including the length and the behavior of link beams (shear/flexural) were studied. An ideal artificial neural network was moreover designed based on the parameters, and experimental equations were proposed to determine the R factor under consecutive earthquakes with acceptable accuracy. The results indicated that the designed buildings for the R factor based on single earthquake could not provide the expected performance when experiencing successive shocks because the R factors caused by consecutive earthquakes was 26% lower than the single case.

Microseismic network sensitivity in case of no seismic activity

Underground human activities, such as mining, shale gas, and oil exploitation, waste-water disposal, or geothermal plants, can cause earthquakes; therefore, they are monitored by local seismic networks. An ideal seismic network has a triangulated grid, with spacing equal twice the minimal depth and no associated industry noise. In real cases, the network sensitivity is biased by stations placed near noisy roads, factories, or in a private garden, none located at optimal nodes. The sensitivity is also a function of the detection algorithm type and setting. The goal of this case study is to suggest a work-flow for network sensitivity calculation in case of no seismic activity. In other words: how small are the earthquakes that such seismic networks would detect if they were present? Such network sensitivity is a function of stations noise level, station-source geometry, and setting of the detection algorithm. A brief theory and work-flow description is followed by two real-case demonstrations from Czech Republic, Europe: first, a proof-test on a well-studied seismically active area of West Bohemia/Vogtland and second, an application to an uprising geothermal project in Litoměřice, where no seismic activity was detected in years of monitoring.

Prediction of transformation temperatures of NiTiZr shape memory alloys using artificial neural network

In this study, detailed analyses have been made to predict the transition temperatures (Ms, Mf, As and Af) and thermal hysteresis of ternary NiTiZr shape memory alloys (SMAs) using an artificial neural network (ANN). The ANN model was built, trained and tested using 35 data points pertaining to Ni-Ti-Zr collected from the literature spanning over 25 years. The alloys that were subjected to age hardening were excluded from the study. In this work, we have used 11 inputs, including the atomic percentage of the constituting elements, i.e. Ni, Ti, and Zr, number of valence electrons, atomic number, valence electron concentration, atomic radius, electronegativity, melting temperature, pseudopotential radius, and atomic mass. By determining the correlation coefficient (R) of different combinations of layers and neurons, an ideal neural network was built. Using these data points, the ANN model was trained and tested in order to produce the output parameters, namely the transition temperatures of the SMAs. The transformation temperatures as predicted by the ANN are compared with those for the predetermined target values. An analysis of the errors involved in predicting the transformation temperatures is also included. The transition temperatures of novel Ni-Ti-Zr alloy compositions were estimated, and their Ms temperatures were set to be substantially above 400 K. The experimentally produced alloy compositions were characterized for their characteristic temperatures of transition using a differential scanning calorimeter so as to validate the outputs obtained.

TNT: A Modular Approach to Traversing Physically Heterogeneous NOCs at Bare-wire Latency

The ideal latency for on-chip network traversal would be the delay incurred from wire traversal alone. Unfortunately, in a realistic modular network, the latency for a packet to traverse the network is significantly higher than this wire delay. The main limiter to achieving lower latency is the modular quantization of network traversal into hops. Beyond this, the physical heterogeneity in real-world systems further complicate the ability to reach ideal wire-only delay. In this work, we propose TNT or Transparent Network Traversal . TNT targets ideal network latency by attempting source to destination network traversal as a single multi-cycle ‘long-hop’, bypassing the quantization effects of intermediate routers via transparent data/information flow. TNT is built in a modular tile-scalable manner via a novel control path performing neighbor-to-neighbor interactions but enabling end-to-end transparent flit traversal. Further, TNT’s fine grained on-the-fly delay tracking allows it to cope with physical NOC heterogeneity across the chip. Analysis on Ligra graph workloads shows that TNT can reduce NOC latency by as much as 43% compared to the state of the art and allows efficiency gains up to 38%. Further, it can achieve more than 3x the benefits of the best/closest alternative research proposal, SMART [ 43 ].