Code Vectors Research Articles

Geochemical performance evaluation of pre-flushing of weak and strong acids during pH-triggered polymer flooding

Implementing pH-triggered polymer is an efficient way to improve the conformance control of hydrocarbon formations. pH-triggered polymer solutions have high mobility in low pH conditions, while due to their high swelling capacity in higher pH values, their viscosity increases with pH increment. Therefore, they can fill high permeability zones/strata and consequently improve the efficiency of enhanced oil recovery. To achieve the desired viscosity of a pH-triggered polymer, the pH of porous media is regulated by implementing an appropriate acid pre-flushing. In this study, we develop a mathematical model to predict the acid pre-flush behavior before the main pH-triggered polymer flooding. Convection, diffusion, and homogeneous/heterogeneous geochemical reactions for a slightly compressible fluid flow in porous media is formulated. This model consists of three non-linear conservation equations (i.e., mass balance, species concentration balance, and Darcy law), which are solved simultaneously in a global implicit approach. We discretize the transport and geochemical reaction terms with a finite difference method by implementing the benefit of the Newton-Raphson method. Moreover, the code vectorization programming approach of the Python language is applied to decrease the computation time. The introduced model is validated against the experimental data available in the literature. This EOR method is mostly implemented on sandstones and there are very limited investigations on the applicability of this method on carbonate rocks. Hence, we simulate the injection of several acids (strong and weak acids) into a carbonate rock under different flow conditions (i.e., different Damkohler numbers). The obtained pH values reveal that weak acids can provide better results for acid pre-flushing in carbonate formations. Therefore, implementing these acids can be a more efficient way to increase the acid residence time as well as to improve the conformance control during a pH-triggered polymer flooding project.

Multi-modal semantic autoencoder for cross-modal retrieval

Cross-modal retrieval has gained much attention in recent years. As the research mainstream, most of existing approaches learn projections for data from different modalities into a common space where data can be compared directly. However, they neglect the preservation of feature and semantic information, so they are unable to obtain satisfactory results as expected. In this paper, we propose a two-stage learning method to learn multi-modal mappings that project multi-modal data to low dimensional embeddings that preserve both feature and semantic information. In the first stage, we combine both low-level feature and high-level semantic information to learn feature-aware semantic code vectors. In the second stage, we use encoder–decoder paradigm to learn projections. The encoder projects feature vectors to code vectors, and the decoder projects code vectors back to feature vectors. The encoder-decoder paradigm guarantees the embeddings to preserve both feature and semantic information. An alternating minimization procedure is developed to solve the multi-modal semantic autoencoder optimization problem. Extensive experiments on three benchmark datasets demonstrate that the proposed method outperforms state-of-the-art cross-modal retrieval methods.

Student cluster competition 2017, Team University of Texas at Austin/Texas State University: Reproducing vectorization of the Tersoff multi-body potential on the Intel Skylake and NVIDIA V100 architectures

This paper satisfies the reproducibility challenge of the Student Cluster Competition at Supercomputing 2017. We attempted to reproduce the results of H\"{o}hnerbach et al. (2016) for an implementation of a vectorized code for the Tersoff multi-body potential kernel of the molecular dynamics code Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS). We investigated accuracy, optimization performance, and scaling with our Intel CPU and NVIDIA GPU based cluster.

High-Fidelity Monocular Face Reconstruction Based on an Unsupervised Model-Based Face Autoencoder.

In this work, we propose a novel model-based deep convolutional autoencoder that addresses the highly challenging problem of reconstructing a 3D human face from a single in-the-wild color image. To this end, we combine a convolutional encoder network with an expert-designed generative model that serves as decoder. The core innovation is the differentiable parametric decoder that encapsulates image formation analytically based on a generative model. Our decoder takes as input a code vector with exactly defined semantic meaning that encodes detailed face pose, shape, expression, skin reflectance, and scene illumination. Due to this new way of combining CNN-based with model-based face reconstruction, the CNN-based encoder learns to extract semantically meaningful parameters from a single monocular input image. For the first time, a CNN encoder and an expert-designed generative model can be trained end-to-end in an unsupervised manner, which renders training on very large (unlabeled) real world datasets feasible. The obtained reconstructions compare favorably to current state-of-the-art approaches in terms of quality and richness of representation. This work is an extended version of [1] , where we additionally present a stochastic vertex sampling technique for faster training of our networks, and moreover, we propose and evaluate analysis-by-synthesis and shape-from-shading refinement approaches to achieve a high-fidelity reconstruction.

The LHCb Software and Computing Upgrade towards LHC Run 3

LHCb is planning major changes for its data processing and analysis workflows for LHC Run 3. Removing the hardware trigger, a software only trigger at 30 MHz will reconstruct events using final alignment and calibration information provided during the triggering phase. These changes pose a major strain on the online software framework which needs to improve significantly. The foreseen changes in the area of the core framework include a re-design of the event scheduling, introduction of concurrent processing, optimizations in processor cache accesses and code vectorization. Furthermore changes in the areas of event model, conditions data and detector description are foreseen. The changes in the data processing workflow will allow an unprecedented amount of signal events to be selected and therefore increase the load on the experiments simulation needs. Several areas of improvement for fast simulation are currently being investigated together with improvements needed in the area of distributed computing. Finally the amount of data stored needs to be reflected in the analysis computing model where individual user analysis on distributed computing resources will become inefficient. This contribution will give an overview of the status of those activities and future plans in the different areas from the perspective of the LHCb computing project.

Easy In Vitro Synthesis of Optimised Functioning Reporter mRNA from Common eGFP Plasmid.

The extensive growth in number and importance of experiments and clinical-aimed techniques based solely or majorly on the activity of RNA strands, e.g. CRSPR/Cas9 and siRNA, has put emphasis on the necessity of standardisation of experiments with RNA. Considering RNA degradation during its handling seems to be a major hindrance in all RNA-based tools, the assessment of its integrity is of utmost importance. Furthermore, evaluating whether the RNA to be transfected is intact requires time-consuming electrophoresis protocol. In view of the RNA lability and the necessity for controlling experiments performed with this molecule, the transfection of a reporter mRNA may be of aid in optimising experiments. Nevertheless, commercial reporter mRNAs are far less available than plasmids for such purpose. Thus, in this work, we aimed at the optimisation of an easily performed protocol to produce a suitable eGFP mRNA. By utilising molecular biology kits customarily employed in molecular biology laboratories working with RNA-based techniques and starting from any eGFP coding vector, we produced four mRNA molecules: (1) eGFP mRNA (non-polyadenylated); (2) Kozak-eGFP mRNA (non-polyadenylated, produced from the Kozak-containing amplicon); (3) eGFP-PolyA mRNA (polyadenylated); (4) Kozak-eGFP-PolyA mRNA (containing both signals, Kozak sequence and poly(A) tail). These mRNA molecules were transfected into HEK 293 FT cells, readily transfectable, and into the MDBK bovine lineage, which has been observed as difficult-to-transfect DNA constructs. eGFP expression could be detected both by flow cytometry and by fluorescence microscopy after transfection with the polyadenylated mRNAs. Upon cytometric analysis, we noted a marked difference among the mRNA groups (p < 0.01), both in fluorescent population percentage and in florescence intensity. We showed here the necessity of the polyadenylation step in order to achieve cell expression of the eGFP observable under fluorescence microscopy. The presence of the Kozak sequence, as a 5' element, seems to augment significantly the level of protein produced upon mRNA transfection. We presented here an easy protocol to allow production of functioning mRNAs from any DNA construct. The molecules produced may aid in the standardisation and controlling most of the RNA-related experiments as well as it gives proper guidance for researchers performing expression of other proteins through mRNA transfection.

Prediction task guided representation learning of medical codes in EHR

There have been rapidly growing applications using machine learning models for predictive analytics in Electronic Health Records (EHR) to improve the quality of hospital services and the efficiency of healthcare resource utilization. A fundamental and crucial step in developing such models is to convert medical codes in EHR to feature vectors. These medical codes are used to represent diagnoses or procedures. Their vector representations have a tremendous impact on the performance of machine learning models. Recently, some researchers have utilized representation learning methods from Natural Language Processing (NLP) to learn vector representations of medical codes. However, most previous approaches are unsupervised, i.e. the generation of medical code vectors is independent from prediction tasks. Thus, the obtained feature vectors may be inappropriate for a specific prediction task. Moreover, unsupervised methods often require a lot of samples to obtain reliable results, but most practical problems have very limited patient samples. In this paper, we develop a new method called Prediction Task Guided Health Record Aggregation (PTGHRA), which aggregates health records guided by prediction tasks, to construct training corpus for various representation learning models. Compared with unsupervised approaches, representation learning models integrated with PTGHRA yield a significant improvement in predictive capability of generated medical code vectors, especially for limited training samples.

Compressed Sensing SAR Imaging Based on Centralized Sparse Representation

Sparse representation based synthetic aperture radar (SAR) imaging approaches have shown their superior performance and great potential in compressed sensing SAR imaging field. However, for many existing approaches, the reconstruction accuracy may be affected by inexact observation and low radar sampling ratio. Inspired by sparse representation works in image restoration, we proposed a novel centralized sparse representation based SAR imaging approach. We assume that the imaging result is the degraded (noisy or inaccurate reconstructed) version of the true scattered field. So, the aim of reconstruction is to make the sparse coding coefficients of imaging result as close as possible to the coefficients of the true scattered field. Under this assumption, a centralized sparsity constraint is added into the reconstruction framework to shrink the difference between the coefficients of imaging result and true scattered field and hence make the sparse coding more accurate. To solve the joint optimization problem between updating sparse coding vectors and SAR imaging, a joint optimization method using L q (0 < q ≤ 1) regularization is proposed. Experimental results have proven to demonstrate the validity of the proposed approach.

Iterative differential quadrature algorithms for modified Burgers equation

PurposeThis paper aims to investigate the features of three vectorized iterative numerical schemes used to simulate the behavior of modified Burgers equation (MBE).Design/methodology/approachTwo of the schemes comprise differential quadrature and finite difference methods, while the third scheme consists of only differential quadrature for the derivative approximations. Proposed schemes are simulated for well-posed problems of MBE having known the analytic solution. The computational complexity of the schemes is examined through monitoring the time taken to complete the simulation. The results are compared with the analytic solution with the help of discrete error norms. Also, the accuracy of the proposed schemes is compared with that of the existing schemes in the literature. Vectorized MATLAB programs of the schemes are used for all investigations.FindingsIt is observed that all the three schemes succeeded in producing a good replication of the exact solution. The results are closer to the analytical solution than the results in the literature. Among the three schemes, the scheme labeled as FDTDQS is found highly accurate and computationally cheaper using fewer grid points. From the vectorized MATLAB programs provided, it is evident that the implementation of the schemes is simple.Originality/valueThis study gives an idea about three numerical schemes for a highly nonlinear problem. This mathematical framework can be adopted to any one-dimensional partial differential equation as well, and the provided program will be helpful to generate more fast and accurate vectorized code in MATLAB.

Codebooks design and performance evaluation based on antenna array response and signal to noise ratio for mmWave communication

ABSTRACTCodebook design is a key point in beam-forming which is necessary for millimetre Wave (mmWave) communication to compensate its huge free space path loss. In beam-forming codebook, the transmitter and the receiver co-operate to select the optimum pre-designed code vectors to satisfy robust, efficient, and reliable transmission link between two pairs. In our study, we propose a new codebook structure based on unique phase states. Based on the Array Factor (AF) and Signal-to-Noise Ratio (SNR), the proposed codebook is compared and evaluated with the uniform weighting codebook, circular array codebook, and IEEE 802.15.3c standard codebook. The results show that the proposed codebook achieves comparable SNR and array directivity with the IEEE 802.15.3c, but higher peak and wide dynamic range of SNR and better array directivity compared with the uniform weighting and circular antenna codebooks. Thus, the proposed codebook is expected to improve signal quality and increase user throughput significantly.

Anonymous Communication Scheme for Wireless Mesh Network Based on Network Coding

Wireless Mesh Network, holding the features of self-organization, strong scalability, low deploymentcost, is considered to be the most competitive technology for broadband wireless access service.However, the security and privacy issues become the bottleneck for WMN’s rapid popularization.Network coding, which can improve both network throughput and anonymity, is adopted for thecommunication security in WMN recently. However, the related network-coding based anonymouscommunication schemes suffer from the efficiency and security issues. In this paper, an efficientanonymous communication scheme for WMN is proposed, which combines the opportunistic routingprotocol and network coding approach. The global coding vector is protected with the lightweightpermutation encryption mechanism. In addition, the security and performance analysis demonstratethat the proposed scheme can provide strong privacy protection for WMN communication with highnetwork throughput.

A Stepwise Analytical Projected Gradient Descent Search for Hyperspectral Unmixing and Its Code Vectorization

We present, in this paper, a new methodology for spectral unmixing, where a vector of fractions, corresponding to a set of endmembers (EMs), is estimated for each pixel in the image. The process first provides an initial estimate of the fraction vector, followed by an iterative procedure that converges to an optimal solution. Specifically, projected gradient descent (PGD) optimization is applied to (a variant of) the spectral angle mapper objective function, so as to significantly reduce the estimation error due to amplitude (i.e., magnitude) variations in EM spectra, caused by the illumination change effect. To improve the computational efficiency of our method over a commonly used gradient descent technique, we have analytically derived the objective function’s gradient and the optimal step size (used in each iteration). To gain further improvement, we have implemented our unmixing module via code vectorization, where the entire process is “folded” into a single loop, and the fractions for all of the pixels are solved simultaneously. We call this new parallel scheme vectorized code PGD unmixing (VPGDU). VPGDU has the advantage of solving (simultaneously) an independent optimization problem per image pixel, exactly as other pixelwise algorithms, but significantly faster. Its performance was compared with the commonly used fully constrained least squares unmixing (FCLSU), the generalized bilinear model (GBM) method for hyperspectral unmixng, and the fast state-of-the-art methods, sparse unmixing by variable splitting and augmented Lagrangian (SUnSAL) and collaborative SUnSAL (CLSUnSAL) based on the alternating direction method of multipliers. Considering all of the prospective EMs of a scene at each pixel (i.e., without a priori knowledge which/how many EMs are actually present in a given pixel), we demonstrate that the accuracy due to VPGDU is considerably higher than that obtained by FCLSU, GBM, SUnSAL, and CLSUnSAL under varying illumination, and is, otherwise, comparable with respect to these methods. However, while our method is significantly faster than FCLSU and GBM, it is slower than SUnSAL and CLSUnSAL by roughly an order of magnitude.

Unsupervised t-Distributed Video Hashing and Its Deep Hashing Extension.

In this paper, a novel unsupervised hashing algorithm, referred to as t-USMVH, and its extension to unsupervised deep hashing, referred to as t-UDH, are proposed to support large-scale video-to-video retrieval. To improve robustness of the unsupervised learning, the t-USMVH combines multiple types of feature representations and effectively fuses them by examining a continuous relevance score based on a Gaussian estimation over pairwise distances, and also a discrete neighbor score based on the cardinality of reciprocal neighbors. To reduce sensitivity to scale changes for mapping objects that are far apart from each other, Student t-distribution is used to estimate the similarity between the relaxed hash code vectors for keyframes. This results in more accurate preservation of the desired unsupervised similarity structure in the hash code space. By adapting the corresponding optimization objective and constructing the hash mapping function via a deep neural network, we develop a robust unsupervised training strategy for a deep hashing network. The efficiency and effectiveness of the proposed methods are evaluated on two public video collections via comparisons against multiple classical and the state-of-the-art methods.

Evaluation of the computational performance of the finite-volume atmospheric model of the IAP/LASG (FAMIL) on a high-performance computer

AbstractHigh computational performance is extremely important for climate system models, especially in ultra-high-resolution model development. In this study, the computational performance of the Finite-volume Atmospheric Model of the IAP/LASG (FAMIL) was comprehensively evaluated on Tianhe-2, which was the world’s top-ranked supercomputer from June 2013 to May 2016. The standardized Atmospheric Model Inter-comparison Project (AMIP) type of experiment was carried out that focused on the computational performance of each node as well as the simulation year per day (SYPD), the running cost speedup, and the scalability of the FAMIL. The results indicated that (1) based on five indexes (CPU usage, percentage of CPU kernel mode that occupies CPU time and of message passing waiting time (CPU_SW), code vectorization (VEC), average of Gflops (Gflops_AVE), and peak of Gflops (Gflops_PK)), FAMIL shows excellent computational performance on every Tianhe-2 computing node; (2) considering SYPD and the cost speedup of ...

Construction of MDS Convolutional Error-Correcting Network Codes Over Cyclic Networks

Recently, the refined singleton bound over acyclic networks was extended to convolutional error-correcting network codes over cyclic networks using extended coding vectors. In this paper, it is shown that constructing an MDS code is equivalent to constructing a multicast code. This is used to develop an algorithm for constructing MDS field-based codes over acyclic networks when the sinks know the topology of the network and the network coding employed at all nodes. A lower bound is given on the size of the field required for the algorithm to be successful. Then this algorithm is extended to construct MDS convolutional error-correcting codes over cyclic networks. The complexity of the proposed algorithm is evaluated.

Supervised neighborhood regularized collaborative representation for face recognition

How to represent a test sample is very crucial for linear representation based classification. The famous sparse representation focuses on employing linear combination of small samples to represent the query sample. However, the local structure and label information of data are neglected. Recently, locality-constrained collaborative representation (LCCR) has been proposed and integrates a kind of locality-constrained term into the collaborative representation scheme. For each test sample, LCCR mainly considers its neighbors to deal with noise and LCCR is robust to various corruptions. However, the nearby samples may not belong to the same class. To deal with this situation, in this paper, we not only utilize the positive effect of neighbors, but also consider the side effect of neighbors. A novel supervised neighborhood regularized collaborative representation (SNRCR) is proposed, which employs the local structure of data and the label information of neighbors to improve the discriminative capability of the coding vector. The objective function of SNRCR obtains the global optimal solution. Many experiments are conducted over six face data sets and the results show that SNRCR outperforms other algorithms in most case, especially when the size of training data is relatively small. We also analyze the differences between SNRCR and LCCR.

End-to-End Feature-Aware Label Space Encoding for Multilabel Classification With Many Classes.

To make the problem of multilabel classification with many classes more tractable, in recent years, academia has seen efforts devoted to performing label space dimension reduction (LSDR). Specifically, LSDR encodes high-dimensional label vectors into low-dimensional code vectors lying in a latent space, so as to train predictive models at much lower costs. With respect to the prediction, it performs classification for any unseen instance by recovering a label vector from its predicted code vector via a decoding process. In this paper, we propose a novel method, namely End-to-End Feature-aware label space Encoding (E2FE), to perform LSDR. Instead of requiring an encoding function like most previous works, E2FE directly learns a code matrix formed by code vectors of the training instances in an end-to-end manner. Another distinct property of E2FE is its feature awareness attributable to the fact that the code matrix is learned by jointly maximizing the recoverability of the label space and the predictability of the latent space. Based on the learned code matrix, E2FE further trains predictive models to map instance features into code vectors, and also learns a linear decoding matrix for efficiently recovering the label vector of any unseen instance from its predicted code vector. Theoretical analyses show that both the code matrix and the linear decoding matrix in E2FE can be efficiently learned. Moreover, similar to previous works, E2FE can be specified to learn an encoding function. And it can also be extended with kernel tricks to handle nonlinear correlations between the feature space and the latent space. Comprehensive experiments conducted on diverse benchmark data sets with many classes show consistent performance gains of E2FE over the state-of-the-art methods.

Relay Selection in Adaptive Buffer-Aided Space-Time Coding with TAS for Cooperative Wireless Networks

In this work, we propose an adaptive buffer-aided space-time coding scheme for cooperative wireless networks. A maximum likelihood receiver and adjustable code vectors are considered subject to a power constraint with an amplify-and forward cooperation strategy. Each relay is equipped with a buffer and is capable of storing the received symbols before forwarding them to the destination. We also present an adaptive relay selection and a new hybrid relay selection protocol, in which the instantaneous signal to noise ratio in each link is calculated and compared at the destination. The transmit antenna selection (TAS) is adopted at the destination where a single transmit antenna at the transmitter and relay that maximizes the instantaneous received signal-to-noise ratio is selected for further transmissions to improve the performance of co-operative network. A stochastic gradient algorithm is then developed to compute the parameters of the adjustable code vector with reduced computational complexity. Simulation results show that the proposed buffer-aided schemes with TAS obtain performance gains over existing schemes.

Experiences of Undergraduate Students Identifying Proteins of Unknown Function As Part of a Teaching Laboratory in a Biochemistry Course

In the laboratory section of Biochemistry II at St. Mary's University, we are performing a series of experiments involving authentic scientific inquiry. The experiments' goal is to identify the functions of a handful of unknown proteins. The curriculum relies heavily on both in silico and in vitro biochemistry techniques. Faculty and students pick the proteins by looking through the PDB list of crystallized proteins that have no known function. Since these proteins have been crystallized, they have already been cloned and purified; we know we are likely to be able to isolate suitable quantities of the proteins for work. We ascertain ahead of time that their coding vectors are available to us from sources such as DNASU or AddGene. We analyze the sequences of potential proteins using internet programs such as Pfam, Dali, BLAST, and HHPred. Good alignment with characterized proteins allows us to make hypotheses for the function of the unknown ones. We use the PyMol plug‐in ProMol to look for arrangements of active site residues common to certain categories of enzymes. For example, we used the triad of serine, histidine, and aspartate from chymotrypsin to identify possible hydrolases. We use the three‐dimensional alignment function of the program Chimera to see whether our unknown proteins resemble the known proteins that are thought to be similar. We overexpress the unknown proteins in E. coli, and purify the enzymes taking advantage of their engineered tags. The spring 2016 class focused on and was successful in broadly identifying four unknown hydrolases. Of these, three were lipases and one was a nuclease. We are extending that success by purifying and assaying additional hydrolases in spring 2017. After every lab in the pilot semester students reported on their experiences, and they presented posters at a University research symposium. The student authors of this poster will either have performed the pilot semester (SM) or be involved in the second round of the curriculum (JB, JB, and EV). They will report on their experiences, their experiments, their findings and whether this innovative teaching lab affected their attitudes toward the inquiry process.Support or Funding InformationThis work is funded by NSF DUE‐1502720.

Speech Recognition in ATMs: Application of Linear Predictive Coding and Support Vector Machines

Speech Recognition in ATMs: Application of Linear Predictive Coding and Support Vector Machines