Tunable Network Research Articles

Robust estimation of shear wave velocity in a carbonate oil reservoir from conventional well logging data using machine learning algorithms

In recent decades, various applications of shear wave velocity have been reported in oil and gas projects. These applications include the determination of lithology, estimation of geomechanical parameters, reservoir fluid detection, etc. The shear wave velocity is usually measured or recorded by DSI (Dipole Shear Sonic imager) instruments. This log is not usually available in all wells of a hydrocarbon field because of the high cost of running this log compared to other conventional logs. However, it is tried to estimate the shear wave velocity from other related logs.The literature contains many empirical rock physics models used to estimate shear wave velocity. Most of them were developed for sandstone lithologies and didn't applicable in carbonate reservoir because of existing some vertical and horizontally complex fracture networks in the carbonate reservoirs that cause a quite different behavior for the shear wave velocity.The main objective of this research is to concentrate on the effect of lithology in the estimation of the target log for a carbonate reservoir. At the earliest step, preprocessing is performed and 455 data points of several conventional well logs, i.g., acoustic, density, neutron porosity, resistivity, gamma-ray, calcite volume, dolomite volume, and water saturation logs have been applied to generate the synthetic shear wave velocity log in a carbonate oil reservoir.First, original data are divided into training, validation, and testing subsets. Different data-driven predictive models were built: Multiple Linear Regression (MLR), Ensemble Learning Method (ELM), and three architectures of Artificial Neural Networks (ANN). The performance of all techniques was compared with each other. Finally, the feed-forward neural network shows the highest accuracy with R-values of 0.99 and 0.96 for the training and testing datasets. Moreover, the network design and tuning parameters of the model are adjusted using the grid search optimization technique to obtain the optimal design of the artificial neural network for estimating the shear wave velocity log. Hence, a deep artificial neural network, i.e., [8–11–15-1] network with bayesian regularization training algorithm and hyperbolic tangent sigmoid transfer function, is proposed to predict the target response in other wells.

Dynamic and Stimuli-Responsive Hydrogel Systems for Controlled Release of Proteins

Background and HypothesisHydrogels have been extensively used as biomaterials for controlled release applications due to their biocompatibility and tunable networks. Stimuli-responsive, or smart, hydrogels are of interest due to their potential for localized release based upon local stimuli such as pH. In addition to smart materials, dynamic hydrogels have been utilized to obtain unique release profiles such as biphasic release curves. We hypothesized that by leveraging dynamic and stimuli-responsive click chemistry we can fabricate hydrogels with unique and controlled release profiles.
 Project MethodsA new reversible addition fragmentation chain transfer (RAFT) polymer known as PADO, was synthesized bearing a ketone functional group allowing for crosslinking with hydrazide bearing motifs to form a pH labile acylhydrazone bond. In addition, polyethylene glycol (PEG) based hydrogels have been extensively used in controlled release applications due to their biocompatibility and antifouling properties. Previously, we have developed a novel conjugation technique for functionalizing norbornene onto hydroxyl-terminated PEG through a cyclic desymmetrization reaction with carbic anhydride. The new polymer, PEGNBCA, was further conjugated with dopamine via amide conjugation to form PEGNBD. Mass loss and protein release analysis were conducted for PADO and PEGNBD hydrogels, respectively.
 ResultsMass loss studies confirmed that PADO hydrogel degraded faster at mildly acidic pH of 6 compared to physiological pH of 7.4. In addition, the rapid hydrolysis of PEGNBD crosslinked via inverse electron demand Diels-Alder reaction demonstrated higher protein release compared to conventional PEGNB hydrogels, which are hydrolytically resistant.
 Conclusion/ImpactThe sensitivity to degradation PADO hydrogels in an acidic environment, and the hydrolytically sensitive PEGNBD hydrogels, can be utilized for targeted therapeutic release. For example, the pH-sensitive hydrogel system can be utilized for therapeutic release into an acidic neoplasm while having minimal release in neutral pH tissues.

GARL-Net: Graph Based Adaptive Regularized Learning Deep Network for Breast Cancer Classification

Across the globe, women suffer from breast cancer fatal disease. It is arising surprisingly due to a lack of awareness among them and the inconvenient reach of diagnostic systems. Many computer-aided diagnostic systems have been developed for the detection of cancer. These systems have quite lower performance, so more accurate diagnosis is the need of the time to save the life of human beings. For large and imbalanced image datasets, efficient learning of the network is very important to detect and classify breast cancer more accurately. In this paper, a graph based adaptive regularized learning of deep network (GARL-Net) is proposed for more accurate breast cancer classification. Transfer learning is used for training the backbone network DenseNet121. furthermore, fine tuning of backbone network is followed by the estimated improved loss function. The improved loss function is actually graph based adaptively regularized complement cross entropy loss. The SoftMax cross entropy in itself is not sufficient to classify image samples accurately, so complement entropy technique is incorporated with the cross-entropy loss to overcome the misclassification issue. Further, by adaptive scaling of regularization term with spatial graph Laplacian basis used to adaptively penalize the complement cross entropy loss for improving the learning of the network. The performance of the proposed method is evaluated using BreakHis and BACH 2018 histopathology image datasets and outperforms the existing state-of-the-art methods and achieved 99.00% of precision, 99.40% of recall, 99.20 % of F-1score, and 99.49% of accuracy for binary classification of breast cancer image samples of BreakHis dataset.

Diabetic Retinopathy Grading From Color Fundus Images: An Autotuned Deep Learning Approach

Diabetic Retinopathy (DR) is a consequence of long-term diabetes which affects the eyes. It causes affliction to the veins in the eyes causing raptures on the retina which can impact vision. If the condition is failed to be detected at an early stage, it can lead to complete vision loss. The conventional diagnosing cycle of DR using fundus images requires very skilled practitioners due to the minute nature of features of the anomalies which can even lead to misdiagnosis and is time consuming. Therefore, devising an automated method for the diagnosis of DR can assist individuals with diabetes to identify symptoms of DR at a very early stage. This research also encompasses the classification of the detected images into its corresponding 3 stages namely No Diabetic Retinopathy (No DR), Non-Proliferative Diabetic Retinopathy (NPDR) and Proliferative Diabetic Retinopathy (PDR) that can greatly aid in monitoring the dynamics of the key features including lesions, hemorrhages, and density of blood vessels. Deep learning algorithms has gotten to become a well-known method that can accomplish a wide variety of classification tasks. However, most of these methods are only efficient in classifying the various stages of DR with low accuracies notably, for the early stages. The devised algorithm in this research uses enriched images processing techniques, automatic hyperparameter tuning and neural network training strategies to provide more emphasis on the minute features for better prediction. The algorithm was tested and compared with modified Resnet50, VGG16, Mobilenetv2, Inceptionv3 and InceptionResnetv2 which gave a classification accuracy of 94.7%, 86.1%, 85.8%, 85.3% and 87% respectively with corresponding detection accuracies of 99.8%, 94%, 94.2%, 94.9% and 98.2% respectively on a test set of 508 images. Using the proposed algorithm, the results indicate Resnet50 based network gave superior performance for both detection and classification tasks.

Model distillation for high-level semantic understanding：a survey

Model distillation for high-level semantic understanding：a survey

Mechanically tunable supramolecular polymer networks with different triblock backbones

AbstractSupramolecular polymer networks (SPNs) have attracted much research attention due to their good performance in the preparation of adaptive materials. Controlling the distribution of supramolecular interactions in the networks is a good strategy to tune the properties of SPNs, but corresponding study is still rare. Herein, we construct two triblock copolymers in ABA and BAB forms to control the supramolecular interaction concentrated in A block, leading to two SPNs with divergent mechanical performances. In specific, the SPN‐1 formed by ABA triblock copolymer displays a slightly and densely distributed hard domain and a compact network, resulting in a decent toughness. While the SPN‐2 composed of BAB triblock copolymer is rigid and less elastic, whose microphase‐separated structures are more obvious, making the network difficult to withstand larger strains. These findings are conducive to deepening the understanding of SPNs and would promote the development of supramolecular block copolymers with emergent functions and applications.

Critical periods in Drosophila neural network development: Importance to network tuning and therapeutic potential.

Critical periods are phases of heightened plasticity that occur during the development of neural networks. Beginning with pioneering work of Hubel and Wiesel, which identified a critical period for the formation of ocular dominance in mammalian visual network connectivity, critical periods have been identified for many circuits, both sensory and motor, and across phyla, suggesting a universal phenomenon. However, a key unanswered question remains why these forms of plasticity are restricted to specific developmental periods rather than being continuously present. The consequence of this temporal restriction is that activity perturbations during critical periods can have lasting and significant functional consequences for mature neural networks. From a developmental perspective, critical period plasticity might enable reproducibly robust network function to emerge from ensembles of cells, whose properties are necessarily variable and fluctuating. Critical periods also offer significant clinical opportunity. Imposed activity perturbation during these periods has shown remarkable beneficial outcomes in a range of animal models of neurological disease including epilepsy. In this review, we spotlight the recent identification of a locomotor critical period in Drosophila larva and describe how studying this model organism, because of its simplified nervous system and an almost complete wired connectome, offers an attractive prospect of understanding how activity during a critical period impacts a neuronal network.

Double scaffold networks regulate edible Pickering emulsion gel for designing thermally actuated 4D printing

3D and 4D printing of emulsion gels can be achieved by controlling the continuous phase and the interface. Edible high internal phase water-in-oil Pickering emulsion gels (PEGs) with a tunable double scaffold network structure are designed and prepared by food-grade phytosterol nanoparticles (PPs). In PEGs, PPs through hydrogen bonding enable binding of emulsion droplets to form the first scaffold network, giving PEGs high viscoelasticity for 3D printing. In the continuous oil phase, palm kernel stearin (PKST) can crystallize forming the second scaffold network of crystals to reinforce 3D printed objects of PEGs. The PEG can be used as a biocompatible template to engineer edible and rigid porous materials with adjustable strength and pore size depending on the degree of curing. 4D printing of PEGs is achieved by the thermal response of the PKST crystal network, leading to the unlimited potential of highly biocompatible PEGs in many applications.

Reentrant rigidity percolation in structurally correlated filamentous networks

Many biological tissues feature a heterogeneous network of fibers whose tensile and bending rigidity contribute substantially to these tissues' elastic properties. Rigidity percolation has emerged as an important paradigm for relating these filamentous tissues' mechanics to the concentrations of their constituents. Past studies have generally considered tuning of networks by spatially homogeneous variation in concentration, while ignoring structural correlation. We introduce here a model in which dilute fiber networks are built in a correlated manner that produces alternating sparse and dense regions. Our simulations indicate that structural correlation consistently allows tissues to attain rigidity with less material. We further find that the percolation threshold varies nonmonotonically with the degree of correlation, such that it decreases with moderate correlation and once more increases for high correlation. We explain the eventual reentrance in the dependence of the rigidity percolation threshold on correlation as the consequence of large, stiff clusters that are too poorly coupled to transmit forces across the network. Our study offers deeper understanding of how spatial heterogeneity may enable tissues to robustly adapt to different mechanical contexts.

Active Control of Plasmonic-Photonic Interactions in a Microbubble Cavity.

Active control of light-matter interactions using nanophotonic structures is critical for new modalities for solar energy production, cavity quantum electrodynamics (QED), and sensing, particularly at the single-particle level, where it underpins the creation of tunable nanophotonic networks. Coupled plasmonic-photonic systems show great promise toward these goals because of their subwavelength spatial confinement and ultrahigh-quality factors inherited from their respective components. Here, we present a microfluidic approach using microbubble whispering-gallery mode cavities to actively control plasmonic-photonic interactions at the single-particle level. By changing the solvent in the interior of the microbubble, control can be exerted on the interior dielectric constant and, thus, on the spatial overlap between the photonic and plasmonic modes. Qualitative agreement between experiment and simulation reveals the competing roles mode overlap and mode volume play in altering coupling strengths.

A Fast and Efficient Beamforming Algorithm Imitating Plant Growth Gene for Phased Array Antenna

In this paper, a new beamforming algorithm for phased array antennas is proposed, the plant growth gene algorithm. The algorithm consists of three steps. Firstly, according to the excitation relation of the array unit before and after the local fine-tuning of the antenna radiation pattern, the model for solving the array unit excitation difference is established. Secondly, the Taylor series expansion is used to solve the model, and the growth model is established based on this, and the beam tuning network is designed to realize the growth model. Finally, based on the growth gene obtained by the neural network algorithm, the growth model is called multiple times for high-precision beamforming. This algorithm converts the complex optimization process of array antenna excitation by the classical optimization algorithm into a simple process of fine-tuning the gain at any angle on the beam to make it grow and approach the target pattern. The growth gene is used to weigh the target angle and gain to achieve beamforming, which greatly reduces the complexity of the algorithm and improves its accuracy of the algorithm. Taking a 1 × 16 linear array as an example, a cosecant square beam pattern with a coverage range of −31° to 31° and a maximum gain direction of 17° is designed using the algorithm proposed in this paper. The experimental results show that the proposed algorithm can easily fine-tune the gain of any angle to achieve precise beamforming. Importantly, the growth genes trained by the algorithm are universal to the phased array antenna with the same topology.

Tunable Neural Networks for Multi-Material Image Formation from Spectral CT Measurements.

Quantitative estimation of multi-material density images is an important goal for Spectral CT imaging. However, material decomposition is a poorly-conditioned nonlinear inverse problem. Maximum-likelihood model-based material decomposition results in very noisy material density image estimates. One increasingly popular strategy for noise reduction is to apply deep neural networks for multi-material image formation. The most common loss function is mean squared error with respect to supervised target images such as ground truth or higher-dose cases. However, we believe that the mean-squared error loss function has several issues for multi-material image formation. In this work, we present a new loss function which includes multiple noise realizations with separate weights on covariance and bias for joint denoising of all material bases. By modulating these weights, it is possible to tune the image quality of neural network output images. To demonstrate our proposed approach, we conducted a simulation of a water/calcium/gadolinium spectral CT imaging scenario using a deep neural network for multi-material image denoising. Our results show that by changing the weights of our proposed loss function, it is possible to control the tradeoff between variance and bias for individual materials as well as the control over the bias coupling between materials.

Nn-TransUNet: An Automatic Deep Learning Pipeline for Heart MRI Segmentation.

Cardiovascular disease (CVD) is a disease with high mortality in modern times. The segmentation task for MRI to extract the related organs for CVD is essential for diagnosis. Currently, a large number of deep learning methods are designed for medical image segmentation tasks. However, the design of segmentation algorithms tends to have more focus on deepening the network architectures and tuning the parameters and hyperparameters manually, which not only leads to a high time and effort consumption, but also causes the problem that the architectures and setting designed for a single task only performs well in a single dataset, but have low performance in other cases. In this paper, nn-TransUNet, an automatic deep learning pipeline for MRI segmentation of the heart is proposed to combine the experiment planning of nnU-net and the network architecture of TransUNet. nn-TransUNet uses vision transformers and convolution layers in the design of the encoder and takes up convolution layers as decoder. With the adaptive preprocessing and network training plan generated by the proposed automatic experiment planning pipeline, nn-TransUNet is able to fulfill the target of medical image segmentation in heart MRI tasks. nn-TransUNet achieved state-of-the-art level in heart MRI segmentation task on Automatic Cardiac Diagnosis Challenge (ACDC) Dataset. It also saves the effort and time to manually tune the parameters and hyperparameters, which can reduce the burden on researchers.

Highly Tunable and Robust Dynamic Polymer Networks via Conjugated–Hindered Urea Bonds

Highly Tunable and Robust Dynamic Polymer Networks via Conjugated–Hindered Urea Bonds

Wind turbine power prediction via deep neural network using hybrid approach

Due to the chaotic nature of wind speed, short-term wind power prediction is a challengeable one. A reliable and accurate wind power prediction model is necessary for the wind turbine industry. Based on their improved ability to cope with complicated nonlinear issues, an increasing number of deep learning-based models are being explored for wind power prediction as artificial intelligence technologies, especially in deep neural networks. In this research, the wind power prediction model is divided into three stages. The first stage is to collect data from wind turbines. The second stage is to apply the optimal tuning of the Deep Neural Network (DNN) to predict the wind power. Here, the hybrid algorithm termedBird Swarm Merged Seagull Optimizer(BSMSO) stipulates DNN’s weight optimization points for wind power prediction; in addition, it reduces the time required for the same. Finally, the efficacy of the proposed BSMSO-DNN prediction model is proved by matching the statistical performance measures as regards toerror metrics with other existing techniques. The simulation results reveal thatthe proposed hybrid modelreduces the prediction errors significantly.

CNN-Hyperparameter Optimization for Diabetic Maculopathy Diagnosis in Optical Coherence Tomography and Fundus Retinography

Diabetic Maculopathy (DM) is considered the most common cause of permanent visual impairment in diabetic patients. The absence of clear pathological symptoms of DM hinders the timely diagnosis and treatment of such a critical condition. Early diagnosis of DM is feasible through eye screening technologies. However, manual inspection of retinography images by eye specialists is a time-consuming routine. Therefore, many deep learning-based computer-aided diagnosis systems have been recently developed for the automatic prognosis of DM in retinal images. Manual tuning of deep learning network’s hyperparameters is a common practice in the literature. However, hyperparameter optimization has shown to be promising in improving the performance of deep learning networks in classifying several diseases. This study investigates the impact of using the Bayesian optimization (BO) algorithm on the classification performance of deep learning networks in detecting DM in retinal images. In this research, we propose two new custom Convolutional Neural Network (CNN) models to detect DM in two distinct types of retinal photography; Optical Coherence Tomography (OCT) and fundus retinography datasets. The Bayesian optimization approach is utilized to determine the optimal architectures of the proposed CNNs and optimize their hyperparameters. The findings of this study reveal the effectiveness of using the Bayesian optimization for fine-tuning the model hyperparameters in improving the performance of the proposed CNNs for the classification of diabetic maculopathy in fundus and OCT images. The pre-trained CNN models of AlexNet, VGG16Net, VGG 19Net, GoogleNet, and ResNet-50 are employed to be compared with the proposed CNN-based models. Statistical analyses, based on a one-way analysis of variance (ANOVA) test, receiver operating characteristic (ROC) curve, and histogram, are performed to confirm the performance of the proposed models.

Comonomer-Tuned Gel Electrolyte Enables Ultralong Cycle Life of Solid-State Lithium Metal Batteries.

Rechargeable lithium metal batteries (LMBs) are considered the "holy grail" of energy storage systems. Unfortunately, uncontrollable dendritic lithium growth inherent in these batteries has prevented their practical applications. The benefits of solid-state electrolyte for LMBs are limited due to the common compromise between ionic conductivity and mechanical property. This work proposes a mechanism for simultaneous improvement in ionic conductivity and mechanical strength of gel polymer electrolyte (GPE) which is based on tunable cross-linked polymer network through adjusting monomer ratios. With increasing bisphenol A ethoxylate dimethacrylate (E2BADMA) and poly(ethylene glycol) diacrylate (PEGDA) mass ratios in GPE precursors, the formed polymer network experienced a composition evolution from a 3D cross-linked mono PEGDA network to triple PEGDA, E2BADMA, and PEGDA/E2BADMA networks and then to dual E2BADMA and PEGDA/E2BADMA networks, accompanied by the increase in both storage modulus (from 6 to 37 MPa) and ionic conductivity (from 0.06 to 0.44 mS cm-1). As a result, the E2BADMA/PEGDA mass ratio of 2:1 facilitates the successful fabrication of a dual-network-supported GPE (PEEPL-12) with a mechanical strength of 37 MPa and superior electrochemical properties (a high ionic conductivity of 0.44 mS cm-1 and a wide electrochemical stability window of 4.85 V vs Li/Li+). Such polymer electrolyte-based symmetric lithium metal batteries delivered a long cycle life (2000 h at 0.1 mA cm-2 and 0.1 mAh cm-2), and the Li|PEEPL-12|LiFePO4 cell delivered a high capacity of 140 mAh g-1 at the 100th cycle at the current density of 0.1 C (1 C = 170 mAh g-1). A more thorough investigation indicated the formation of a stable solid electrolyte interphase layer on a lithium metal anode. These extraordinary features open up a venue for fabrication of advanced polymer electrolyte for long-cycle-life lithium metal batteries.

A prior-guided deep network for real image denoising and its applications

Recently, the popularity of deep learning in computer vision has brought about a rapid development of deep convolutional neural network (CNN)-based image denoising algorithms. Most of the existing methods have shown their superiority in removing additive white Gaussian noise (AWGN) with a specific noise level, but have limited performance in handling much more complicated real noise. To address this problem, we develop a novel prior-guided dynamic tunable network (PDTNet) for real image denoising. Firstly, we break up the image denoising optimization problem into noise estimation and image reconstruction sub-problems, and employ the inference process to guide the architecture design of PDTNet. Then, we exploit an internal and external dual modulation (IEDM) scheme to achieve real image denoising. Specifically, a designed global spatial and channel attention (GSCA) is embedded in the external estimator and internal stacked dynamic residual blocks (DRBs) to extract global features from the noise prior and iterative image features, respectively. Next, a dynamic weight generator block (DWGB) is leveraged to adaptively combine external and internal features at each DRB. In addition, we also analyze a realistic real noise model from the physical perspective to generate synthetic noisy images for model training. Experimental results show the superiority of PDTNet over state-of-the-arts both quantitatively and visually and its applications on different networks and tasks.

Reconstituting and Characterizing Actin-Microtubule Composites with Tunable Motor-Driven Dynamics and Mechanics.

The composite cytoskeleton, comprising interacting networks of semiflexible actin filaments and rigid microtubules, restructures and generates forces using motor proteins such as myosin II and kinesin to drive key processes such as migration, cytokinesis, adhesion, and mechanosensing. While actin-microtubule interactions are key to the cytoskeleton's versatility and adaptability, an understanding of their interplay with myosin and kinesin activity is still nascent. This work describes how to engineer tunable three-dimensional composite networks of co-entangled actin filaments and microtubules that undergo active restructuring and ballistic motion, driven by myosin II and kinesin motors, and are tuned by the relative concentrations of actin, microtubules, motor proteins, and passive crosslinkers. Protocols for fluorescence labeling of the microtubules and actin filaments to most effectively visualize composite restructuring and motion using multi-spectral confocal imaging are also detailed. Finally, the results of data analysis methods that can be used to quantitatively characterize non-equilibrium structure, dynamics, and mechanics are presented. Recreating and investigating this tunable biomimetic platform provides valuable insight into how coupled motor activity, composite mechanics, and filament dynamics can lead to myriad cellular processes from mitosis to polarization to mechano-sensation.

Novel Gysel power dividers with tunable power dividing ratio and central frequency

Novel power dividers based on Gysel architecture with tunable power dividing ratio (TPDR), tunable central frequency (TCF), perfect port matching and isolation are proposed. The regulatory mechanism of traditional Gysel Power Divider (GPD) is studied, and a novel electrical tunable method is presented to tune the power divider ratio by only the electrical length of the GPD section, which simplifies the design of tunable power dividers. Tunable T-type networks are used to tune the power dividing ratio and central frequency by tuning this electrical length parameter, and the TPDR range can be pre-defined by setting the characteristic impedance of the transmission lines in the tunable power divider. The tunable mechanism of the circuits was theoretical studied. Two power dividers operating at 950 MHz with TPDR of 0.3:1 ∼ 35.9:1 and 1.1:1 ∼ 35.4:1, and a GPD with TCF in 1.1 GHz ∼ 1.6 GHz were designed and fabricated. All the measured results are in good agreement with the simulation.