Large Number Of Inputs Research Articles

Advanced Methodology for Uncertainty Propagation in Computer Experiments with Large Number of Inputs

In the framework of the estimation of safety margins in nuclear accident analysis, a quantitative assessment of the uncertainties tainting the results of computer simulations is essential. Accurate uncertainty propagation (estimation of high probabilities or quantiles) and quantitative sensitivity analysis may call for several thousand code simulations. Complex computer codes, as the ones used in thermal-hydraulic accident scenario simulations, are often too CPU-time expensive to be directly used to perform these studies. A solution consists in replacing the computer model by a CPU-inexpensive mathematical function, called a metamodel, built from a reduced number of code simulations. However, in case of high-dimensional experiments (with typically several tens of inputs), the metamodel building process remains difficult. To face this limitation, we propose a methodology which combines several advanced statistical tools: initial space-filling design, screening to identify the noninfluential inputs, and Gaussian process (Gp) metamodel building with the group of influential inputs as explanatory variables. The residual effect of the group of noninfluential inputs is captured by another Gp metamodel. Then, the resulting joint Gp metamodel is used to accurately estimate Sobol’ sensitivity indices and high quantiles (here 95% quantile). The efficiency of the methodology to deal with a large number of inputs and reduce the calculation budget is illustrated on a thermal-hydraulic calculation case simulating with the CATHARE2 code a loss-of-coolant accident scenario in a pressurized water reactor. A predictive Gp metamodel is built with only a few hundred code simulations which allows the calculation of the Sobol’ sensitivity indices. This Gp also provides a more accurate estimation of the 95% quantile and associated confidence interval than the empirical approach, at equal calculation budget. Moreover, on this test case, the joint Gp approach outperforms the simple Gp.

A 4K-Input High-Speed Winner-Take-All (WTA) Circuit with Single-Winner Selection for Change-Driven Vision Sensors

Winner-Take-All (WTA) circuits play an important role in applications where a single element must be selected according to its relevance. They have been successfully applied in neural networks and vision sensors. These applications usually require a large number of inputs for the WTA circuit, especially for vision applications where thousands to millions of pixels may compete to be selected. WTA circuits usually exhibit poor response-time scaling with the number of competitors, and most of the current WTA implementations are designed to work with less than 100 inputs. Another problem related to the large number of inputs is the difficulty to select just one winner, since many competitors may have differences below the WTA resolution. In this paper, a WTA circuit is presented that handles more than four thousand inputs, to our best knowledge the hitherto largest WTA, with response times below the microsecond, and with a guaranty of just a single winner selection. This performance is obtained by the combination of a standard analog WTA circuit and a fast digital single-winner selector with almost no size penalty. This WTA circuit has been successfully employed in the fabrication of a Selective Change-Driven Vision Sensor based on 180 nm CMOS technology. Both simulated and experimental results are presented in the paper, showing that a single pixel event can be selected in just 560 ns, and a multipixel pixel event can be processed in 100 μs. Similar results with a conventional approach would require a camera working at more than 1 Mfps for the single-pixel event detection, and 10 kfps for the whole multipixel event to be processed.

Calibration of SWAT and Two Data-Driven Models for a Data-Scarce Mountainous Headwater in Semi-Arid Konya Closed Basin

Hydrologic models are important tools for the successful management of water resources. In this study, a semi-distributed soil and water assessment tool (SWAT) model is used to simulate streamflow at the headwater of Çarşamba River, located at the Konya Closed Basin, Turkey. For that, first a sequential uncertainty fitting-2 (SUFI-2) algorithm is employed to calibrate the SWAT model. The SWAT model results are also compared with the results of the radial-based neural network (RBNN) and support vector machines (SVM). The SWAT model performed well at the calibration stage i.e., determination coefficient (R2) = 0.787 and Nash–Sutcliffe efficiency coefficient (NSE) = 0.779, and relatively lower values at the validation stage i.e., R2 = 0.508 and NSE = 0.502. Besides, the data-driven models were more successful than the SWAT model. Obviously, the physically-based SWAT model offers significant advantages such as performing a spatial analysis of the results, creating a streamflow model taking into account the environmental impacts. Also, we show that SWAT offers the ability to produce consistent solutions under varying scenarios whereas it requires a large number of inputs as compared to the data-driven models.

Intelligent and Secure Content-Based Image Retrieval for Mobile Users

With the tremendous growth of smart mobile devices, the Content-Based Image Retrieval (CBIR) becomes popular and has great market potentials. Secure image retrieval has attracted considerable interests recently due to users' security concerns. However, it still suffers from the challenges of relieving mobile devices of excessive computation burdens, such as data encryption, feature extraction, and image similarity scoring. In this paper, we propose and implement an IND-CPA secure CBIR framework that performs image retrieval on the cloud without the user's constant interaction. A pre-trained deep CNN model, i.e., VGG-16, is used to extract the deep features of an image. The information about the neural network is strictly concealed by utilizing the lattice-based homomorphic scheme. We implement a real number computation mechanism and a divide-and-conquer CNN evaluation protocol to enable our framework to securely and efficiently evaluate the deep CNN with a large number of inputs. We further propose a secure image similarity scoring protocol, which enables the cloud servers to compare two images without knowing any information about their deep features. The comprehensive experimental results show that our framework is efficient and accurate.

A Hybrid Artificial Neural Network Model for Option Pricing

In the absence of a well-defined input selection technique associated with the pure ANN models, Option pricing using pure ANN models while relaxing the assumption of constant volatility remains a challenge. The conservative drill espoused has been to make allowances for a large number of input lags with the confidence that the ability of ANN to integrate suppleness and redundancy generates a more robust model. This is to say that the nonexistence of input selection criteria notwithstanding, the models have been developed without due consideration to the effect that the choice of input selection technique would have on model complexity, learning difficulty and performance measures. In this study, we deviate from the conventional techniques applied by the pure ANN option price models and adopt the hybrid model in which the volatility component is handled using some celebrated time series models, with speci?city to the ANN-GJR-GARCH model - a hybrid of the ANN and a time series hybrid. The hybrid ANN option pricing model is then framed and tested with the forecasts of the ANN-GJR-GARCH model as a volatility input alongside two other inputs - time to maturity and moneyness. Finally, we compare the performance of the hybrid model developed with that of a pure ANN model. Results indicate that the hybrid model outperforms the pure ANN model not only in forecasting but also in the training time and model complexity.

Algebraic Properties of Cores of Generalized Neurofunctions

This paper considers generalized neural elements and identifies the conditions for implementation of functions of algebra of logic from these elements. We introduce the concept of a modified core of Boolean functions with respect to the system of characters of a group on which functions of algebra of logic are given. The criteria of belonging these functions to the class of generalized neurofunctions are provided. The algebraic structure of cores of Boolean neurofunctions is studied. On the basis of properties of tolerance matrices, a number of necessary conditions for implementation of Boolean functions by one generalized neural element are obtained. The obtained results allow to develop efficient methods of synthesis of integer-valued generalized neural elements with a large number of inputs that can be successfully applied in solving information compression and transmission problems, as well as discrete signal recognition problems.

Section-Based Approach Optimizes Unconventional-Reservoir-Fracture Spacing

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 186107, “Optimization of Spacing and Penetration Ratio for Infinite-Conductivity Fractures in Unconventional Reservoirs: A Section-Based Approach,” by S. Liu, SPE, and P.P. Valko, SPE, Texas A&M University. The paper has been peer reviewed and published in the December 2017 SPE Journal. In this paper, the authors consider the development plan of shale gas or tight oil with multiple multistage fractured laterals in a large square drainage area that they call a “section” (usually 640 acres in the US). A section-based optimization of the fracture array is proposed to provide some general statements regarding spacing of wells and fractures. The approach is dependent on a reliable and efficient productivity-index (PI) calculation for the boundary-dominated state (BDS). The information derived from the section-based optimization method and the efficient and reliable algorithm for PI calculation should help the design of multistage fracturing in shale-gas or ultralow-permeability oil formations. Introduction During the last decade, multiple studies have been conducted on various aspects of fracture design. Most of these use a numerical simulation that requires a large number of inputs, which, at the early stages of development, may be in-accessible or burdened by substantial uncertainty. Many of these works use quite sophisticated nonlinear search algorithms. In this work, however, only that part of the problem that can be addressed in rather general terms is considered. Therefore, simplifying assumptions are made to a point at which numerical results can be delivered. The results will provide some insight and useful bounds on achievable productivity. The methodology is dependent on an accurate numerical solution of a well-defined eigenvalue problem and, as such, can provide reference values for various research fields relying heavily on the diffusivity equation. In the complete paper, a method that combines the finite-element method (FEM) with Richardson extrapolation is introduced to calculate the single-fracture PI both for constant-pressure and constant-rate wellbore conditions. Then, with the reliable PI for a single fracture, a convenient section-based optimization is conducted involving only two integer decision variables: the number of columns and the number of rows of the fracture array. Because the constant-pressure wellbore condition is more realistic, the overall PI of the section in the BDS flow regime is considered as the objective function. Two parameters are suggested as the main decisive factors: the dimensionless total fracture length and the feasible range of fracture half-length. In the present approach, the actual optimization takes a simple form of enumeration of a limited number of discrete cases; therefore, advanced nonlinear search algorithms are not involved. Those algorithms become irreplaceable when petrophysical and economic details are incorporated and the structure of the objective function loses its transparent nature, but such situations are beyond the focus of the present work.

Sensitivity of M-E PDG Level 2 and 3 Inputs Using Statistical Analysis Techniques for New England States

Implementation of the Mechanistic-Empirical Pavement Design Guide (M-E PDG) by state highway design agencies necessitates transition from their currently followed pavement design practices along with additional data collection on the large number of inputs incorporated in the M-E PDG. This paper presents a data collection methodology that is representative of a real pavement section instead of a model pavement section or obtaining data from knowledge of engineering principles and experience. A procedure to design field studies is developed such that state highway design specifications can be adjusted if needed and transformed into M-E design guidelines to aid the implementation process. Statistical analysis of the predicted performance data is conducted to assess the sensitivity of Level 2 and 3 inputs on pavement distresses. A roadmap for documentation of design data is proposed based on case-studies from two states, New Hampshire and Connecticut. Input parameters considered to critically affect pavement distresses are identified from previous research results and literature review. Data sources for the corresponding M-E PDG inputs were identified and values for input parameters were selected based on variations that practically occur during pavement design, including tolerances on material properties. Analysis of variance was conducted on the predicted performance data. Percentage variation in the predicted distresses explained by each input variable was calculated and sensitivity levels were determined. Correlation ratio η^2 was used due to the deterministic nature of M-E PDG output. Independent data collection recommendations were made for both states on the basis of the study.

Enhanced Morris method for global sensitivity analysis: good proxy of Sobol’ index

Global sensitivity analysis (GSA) aims at quantifying the effects of inputs on the output response globally. GSA is useful for identifying a few important inputs from a model with large number of inputs, which is critical for structural design and optimization. The method of Sobol’ and the Morris method are two popular GSA techniques, and they have been widely used in many areas of science and engineering. It was proved that the Morris index is a good proxy of the method of Sobol’ in some papers. However, some of the quantitative relationships between Morris index and Sobol’ index are established by only considering the input with standard uniform distribution. When the input does not follow standard uniform distribution, some relationships are no longer valid. Therefore, an enhanced Morris method is developed as a better proxy of the Sobol’ index for the input with arbitrary distribution, and it does not increase the model evaluations compared with the original Morris method. Test examples show the performance of the approximation and its usefulness in practice.

Single-shot and lensless complex-amplitude imaging with incoherent light based on machine learning

We present a single-shot incoherent light imaging method for simultaneously observing both amplitude and phase without any imaging optics, based on machine learning. In the proposed method, an object with a complex-amplitude field is illuminated with incoherent light and is captured by an image sensor with or without a coded aperture. The complex-amplitude field of the object is reconstructed from a single captured image using a state-of-the-art deep convolutional neural network, which is trained with a large number of input and output pairs. In experimental demonstrations, the proposed method was verified with a handwritten character database, and the effect of a coded aperture printed on an overhead projector film in the reconstruction was examined. Our method has advantages over conventional wavefront sensing techniques using incoherent light, namely simplification of the optical hardware and improved measurement speed. This study shows the importance and practical impact of machine learning techniques in various fields of optical sensing.

Asymmetry in indegree and outdegree distributions of gene regulatory networks arising from dynamical robustness.

Although outdegree distributions of gene regulatory networks have scale-free characteristics similar to other biological networks, indegree distributions have single-scale characteristics with significantly lower variance than that of outdegree distributions. In this study, we mathematically explain that such asymmetric characteristics arise from dynamical robustness, which is the property of maintaining an equilibrium state of gene expressions against inevitable perturbations to the networks, such as gene dysfunction and mutation of promoters. We reveal that the expression of a single gene is robust to a perturbation for a large number of inputs and a small number of outputs. Applying these results to the networks, we also show that an equilibrium state of the networks is robust if the variance of the indegree distribution is low (i.e., single-scale characteristics) and that of the outdegree distribution is high (i.e., scale-free characteristics). These asymmetric characteristics are conserved across a wide range of species, from bacteria to humans.

S-box-based random number generation for stochastic computing

Stochastic circuits (SCs) offer tremendous area and power-consumption benefits at the expense of computational inaccuracies. They require random number sources (RNSs) to implement stochastic number generators (SNGs) for all of their inputs. It is common for an SC to have a large number of primary and auxiliary inputs. Often the associated SNGs take up as much as 80% of the entire circuit area, so sharing RNSs is a very important goal in stochastic computing. Such sharing often leads to large correlation errors that have to be resolved via costly decorrelation methods. Linear feedback shift registers (LFSRs) are typically used as RNSs. However, we show that their deterministic and linear behavior can interfere with commonly used decorrelation methods, causing systematic computation errors, and limiting the possibilities of sharing LFSRs between SNGs. We therefore propose a novel pseudo-random number generator SBoNG for stochastic circuits that combines an LFSR with a non-linear S-box function. An SBoNG does not interfere with decorrelation and can be shared efficiently by multiple SNGs. Consequently, SBoNGs scale very well in SCs with large numbers of inputs.

A method for predicting dissolved oxygen in aquaculture water in an aquaponics system

Aquaponics systems combine vegetable cultivation with fish culture (Rao et al., 2017). Several environmental factors influence water quality, among which dissolved oxygen is the most critical. The dissolved oxygen-related processes are nonlinear and unstable and are characterized by considerable time delays and time variances. In an aquaponics system, the dissolved oxygen content is affected by changes in the aquaculture water and in the greenhouse climate. Understanding the changes in dissolved oxygen and ensuring its stability requires effective prediction approaches and corrective measures. Traditional prediction methods are characterized by low stability, low accuracy and poor timeliness. A dissolved oxygen prediction model based on fuzzy neural networks is proposed in this paper. With a large number of inputs, the fuzzy network has high dimensonality, a complex model structure, and low precision, among other limitations. A genetic algorithm can be used to optimize the centre and width of the fuzzy neural network’s middle layer and determine the optimal parameter combination, thus improving the efficiency and predictive accuracy of the model (Liu et al., 2014). The results show that a fuzzy neural network optimized using a genetic algorithm is more stable, more accurate, and more suitable for predicting dissolved oxygen than a fuzzy neural network and backpropagation neural network in an aquaponics system. Predicting dissolved oxygen is critical to the stability of an aquaponics system.

Optimality and identification of dynamic models in systems biology: an inverse optimal control framework.

Optimality principles have been used to explain many biological processes and systems. However, the functions being optimized are in general unknown a priori. Here we present an inverse optimal control framework for modeling dynamics in systems biology. The objective is to identify the underlying optimality principle from observed time-series data and simultaneously estimate unmeasured time-dependent inputs and time-invariant model parameters. As a special case, we also consider the problem of optimal simultaneous estimation of inputs and parameters from noisy data. After presenting a general statement of the inverse optimal control problem, and discussing special cases of interest, we outline numerical strategies which are scalable and robust. We discuss the existence, relevance and implications of identifiability issues in the above problems. We present a robust computational approach based on regularized cost functions and the use of suitable direct numerical methods based on the control-vector parameterization approach. To avoid convergence to local solutions, we make use of hybrid global-local methods. We illustrate the performance and capabilities of this approach with several challenging case studies, including simulated and real data. We pay particular attention to the computational scalability of our approach (with the objective of considering large numbers of inputs and states). We provide a software implementation of both the methods and the case studies. The code used to obtain the results reported here is available at https://zenodo.org/record/1009541. Supplementary data are available at Bioinformatics online.

On-line monitoring of a pilot flotation rougher circuit by using PCA models

Flotation is a very complex process, described by a large number of input and operating variables, frequently high correlated and collinear. Therefore, it is a great challenge to find out which combination of those variables are driving the process out of its target, given by concentrate grade and process recovery. Since most variables are measured with significant error, the on-line identification of which measurement need to be recalibrate is an issue, in order to assure good control decisions. Multivariate statistical projection method, and particularly Principal Component Analysis, can be effectively used to obtain linear models between latent variables directly from operating data. In this work, PCA models have been obtained and used to develop an on-line monitoring system on a pilot rougher circuit. The main experimental results and work to be done are discussed here.

Robust Coplanar Bus with Unique Modular Design for Quantum-dot Cellular Automata

Objective: The proposed designs have been implemented in order to exploit the properties of Quantum Dot Cellular Automata majority voting to consent a modular robust crossing of wires. Methods/Statistical Analysis: Three structures namely 4x4, 5x5 and 8x8 coplanar buses with unique modular design in quantum dot cellular automata have been presented and the operation of these structures has been tested by QCA Designer Simulation Tool. These unique structures present a modus operandi for designing the modified robust coplanar bus with unique modular approach in nanotechnology. Findings: The proposed modular coplanar structures allow number of coplanar wires to cross without signal degradation and signal loss until information propagates the output. Compared to the reported literature, the proposed design has following advantages; • Unique modular approach in which wires can be intersected on the single layer without any intrusion or crosstalk. • High output polarity, fault tolerant capabilities, modular approach and robustness which makes it more efficient without any interference; • Use of sequential clock which makes it easier to understand and manufacture; • The signal and its counterpart can be simply taken out together due to the alternating polarization. • Proper use of clocking phase pairs in a traversed manner in order to permit the accurate propagation of signals. • Cells are evenly disseminated into the clocking zones which are implemented that the unusable spaces can be as small as feasible. • It has been concluded that it is easy to increase the number of inputs by adding the suitable layers with proposed unique modular approach. Application/Improvements: Starting from lowest number of inputs, it is easy to extend the proposed designs to large and then larger number of inputs by adding suitable layers which will prove efficient to tolerate various faults during the manufacturing process.

A risk-based approach for developing standards for irrigation with reclaimed water

A generalised quantitative risk assessment (QRA) is developed to assess the potential harm to human health resulting from irrigation with reclaimed water. The QRA is conducted as a backward calculation starting from a pre-defined acceptable risk level at the receptor point (defined as an annual infection risk of 10−4 for pathogens and by reference doses (RfD) for chemical hazards) and results in an estimate of the corresponding acceptable concentration levels of the given hazards in the effluent. In this way the QRA is designed to inform the level of water treatment required to achieve an acceptable risk level and help establish reclaimed water quality standards. The QRA considers the exposure of human receptors to microbial and chemical hazards in the effluent through various exposure pathways and routes depending on the specific irrigation scenario. By considering multiple pathways and routes, a number of key aspects relevant to estimating human exposure to recycled water can be accounted for, including irrigation and crop handling practices (e.g., non-edible vs edible, spray vs. drip, withholding time) and volumes consumed (directly vs indirectly). The QRA relies on a large number of inputs, many of which were found to be highly uncertain. A possibilistic approach, based on fuzzy set theory, was used to propagate the uncertain input values through the QRA model to estimate the possible range of hazard concentrations that are deemed acceptable/safe for reclaimed water irrigation. Two scenarios were considered: amenity irrigation and irrigation of ready-to-eat food crops, and calculations were carried out for six example hazards (norovirus, Cryptosporidium, cadmium, lead, PCB118 and naphthalene) and using UK-specific input values. The human health risks associated with using reclaimed water for amenity irrigation were overall deemed low, i.e. the calculated acceptable concentration levels for most of the selected hazards were generally far greater than levels typically measured in effluent from wastewater treatment plants; however the predicted acceptable concentration levels for norovirus and Cryptosporidium suggested that disinfection by UV may be required before use. It was found that stricter concentration standards were required for hazards that are more strongly bound to soil and/or are more toxic/infectious. It was also found that measures that reduce the amount of effluent directly ingested by the receptor would significantly reduce the risks (by up to 2 orders of magnitude for the two pathogens). The results for the food crop irrigation scenario showed that stricter concentration standards are required to ensure the effluent is safe to use. For pathogens, the dominant exposure route was found to be ingestion of effluent captured on the surface of the crops indicating that risks could be significantly reduced by restricting irrigation to the non-edible parts of the crop. The results also showed that the exposure to some organic compounds and heavy metals through plant uptake and attached soil particles could be high and possibly pose unacceptable risk to human health. For both scenarios, we show that the predicted acceptable concentration levels are associated with large uncertainty and discuss the implications this has for defining quality standards and how the uncertainty can be reduced.

Using of method of replacement of input variables in microprogram finite-state machine with datapath of transitions

The object of research is a microprogram finite-state machine with datapath of transitions. In its structure, the input signals are fed to the input of the block that forms the codes for the transition operations. A large number of inputs of this unit does not allow to realize it in the basis of memory blocks. The only available basis is the basis of LUT elements of FPGA or similar.To reduce the number of LUT elements used by the circuit of FSM, it is proposed to use the known method of replacement of input variables. Its application leads to the fact that an additional block is added into the structure of the FSM, which converts the input signals of the FSM into special intermediate signals, the number of which is much less than the number of input signals. This leads to a decrease in the input signals of the block that forms the codes of the transition operations, and makes it possible to synthesize it in the basis of the memory blocks without the use of LUT elements.An additional unit that converts input signals into intermediate ones can be synthesized in the basis of multiplexers, which are standard functional blocks of modern FPGAs. This makes it possible to use a smaller number of LUT elements in the structure of the FSM with replacement of input variables than in the prototype structure. Saved in this way LUT elements can be used to implement other units of the FSM or elements of projected computing system.The proposed approach contributes to the saving of hardware amount in the logical circuit of the microprogram finite-state machine with datapath of transitions. This makes this structure more preferable in comparison with the prototype structure from the point of view of hardware amount affecting the final cost of the control unit and the computing system as a whole.

Qutrit witness from the Grothendieck constant of order four

In this paper, we prove that $K_G(3)<K_G(4)$, where $K_G(d)$ denotes the Grothendieck constant of order $d$. To this end, we use a branch-and-bound algorithm commonly used in the solution of NP-hard problems. It has recently been proven that $K_G(3)\le 1.4644$. Here we prove that $K_G(4)\ge 1.4841$, which has implications for device-independent witnessing dimensions greater than two. Furthermore, the algorithm with some modifications may find applications in various black-box quantum information tasks with large number of inputs and outputs.

MIMO Switched-Capacitor DC–DC Converters Using Only Parasitic Capacitances Through Scalable Parasitic Charge Redistribution

This paper presents a multiple-input multiple-output (MIMO) switched-capacitor (SC) dc–dc converter that only uses the parasitic capacitance already present in fully integrated SC power converters to generate multiple dc voltages. When used in an SC converter together with the scalable parasitic charge redistribution technique, the presented MIMO converter provides additional voltage rails, which can be used to power gate drivers or control blocks without any area overhead. Moreover, because the proposed converter only makes use of elements, which are already present in fully integrated SC converters, only conductive losses are introduced. This means that, for low output powers, efficiencies arbitrarily close to 100% can be achieved. The presented type of converter is characterized using an MIMO model, which is, in turn, used to prove the efficiency of the converter compared with regular SC MIMO converters, particularly for a large number of inputs or outputs. Measurements verify the basic working principle of the presented converter, demonstrating a peak efficiency of 98.9% and output powers sufficient to power internal converter blocks.