Multivariate calibration of non-replicated measurements for the factored noise model

The accuracy of a multivariate calibration (MVC) model for relating concentrations of multicomponent mixtures to their spectral measurements depends on effective handling of errors in the measured data. For the case when error variances vary along only one mode (either along mixtures or along wavelengths), a method to estimate the error variances simultaneously along with the spectral subspace was developed by Narasimhan and Shah (Control Engineering Practice, 16, (2008), 146–155). This method was exploited by Bhatt et al. (Chemom. Intell. Lab. Syst., 85, (2007), 70–81) to develop an iterative principal component regression (IPCR) MVC model, which was shown to be more accurate than models developed using PCR. In this work, the IPCR method is extended to deal with measurement errors whose variances vary along both modes, by using a factored noise model. As a first step, an iterative procedure is developed to estimate the error variance factors along with the spectral subspace, which is subsequently used in developing the regression model. Using simulated and experimental data, it is shown that the quality of the MVC model developed using the proposed method is better than that obtained using PCR, and is as good as the model obtained using Maximum Likelihood PCR, which requires knowledge of the error variances. For dealing with large data sets, a sub-optimal approach is also proposed for estimating the large number of error variances.     

Adversarial Active Learning for Named Entity Recognition in Cybersecurity

Owing to the continuous barrage of cyber threats, there is a massive amount of cyber threat intelligence. However, a great deal of cyber threat intelligence come from textual sources. For analysis of cyber threat intelligence, many security analysts rely on cumbersome and time-consuming manual efforts. Cybersecurity knowledge graph plays a significant role in automatics analysis of cyber threat intelligence. As the foundation for constructing cybersecurity knowledge graph, named entity recognition (NER) is required for identifying critical threat-related elements from textual cyber threat intelligence. Recently, deep neural network-based models have attained very good results in NER. However, the performance of these models relies heavily on the amount of labeled data. Since labeled data in cybersecurity is scarce, in this paper, we propose an adversarial active learning framework to effectively select the informative samples for further annotation. In addition, leveraging the long short-term memory (LSTM) network and the bidirectional LSTM (BiLSTM) network, we propose a novel NER model by introducing a dynamic attention mechanism into the BiLSTM-LSTM encoderdecoder. With the selected informative samples annotated, the proposed NER model is retrained. As a result, the performance of the NER model is incrementally enhanced with low labeling cost. Experimental results show the effectiveness of the proposed method.     

Modeling of quantitative relationships between physicochemical properties of active pharmaceutical ingredients and tensile strength of tablets using a boosted tree

Objectives: The aim of this study was to explore the potential of boosted tree (BT) to develop a correlation model between active pharmaceutical ingredient (API) characteristics and a tensile strength (TS) of tablets as critical quality attributes.

Methods: First, we evaluated 81 kinds of API characteristics, such as particle size distribution, bulk density, tapped density, Hausner ratio, moisture content, elastic recovery, molecular weight, and partition coefficient. Next, we prepared tablets containing 50% API, 49% microcrystalline cellulose, and 1% magnesium stearate using direct compression at 6, 8, and 10?kN, and measured TS. Then, we applied BT to our dataset to develop a correlation model. Finally, the constructed BT model was validated using k-fold cross-validation.

Results: Results showed that the BT model achieved high-performance statistics, whereas multiple regression analysis resulted in poor estimations. Sensitivity analysis of the BT model revealed that diameter of powder particles at the 10th percentile of the cumulative percentage size distribution was the most crucial factor for TS. In addition, the influences of moisture content, partition coefficients, and modal diameter were appreciably meaningful factors.

Conclusions: This study demonstrates that BT model could provide comprehensive understanding of the latent structure underlying APIs and TS of tablets.     

Task assignment in loosely‐coupled multiprocessor systems

Abstract

In loosely‐coupled multiprocessor systems, a parallel program has its modules distributedly assigned among the processors. The assignment policy is to minimize interprocessor communication cost and to balance the workload of processors. However, there exists conflict between these two criteria and a compromise must be made to obtain an optimal solution. In this paper, we propose a new task assignment model for distributed computing systems, and for solving the assignment problem based on partitioning graphs. The problem of finding an optimal solution had been shown to be in the class of NP‐complete. For the sake of computation efficiency, we propose some heuristics for obtaining suboptimal solutions.     

18O-tracer diffusion along nanoscaled Sc2O3/yttria stabilized zirconia (YSZ) multilayers: on the influence of strain

Abstract

The oxygen tracer diffusion coefficient describing transport along nano-/microscaled YSZ/Sc₂O₃ multilayers as a function of the thick­ness of the ion-conducting YSZ layers has been measured by isotope exchange depth profiling (IEDP), using secondary ion mass spec­trometry (SIMS). The multilayer samples were prepared by pulsed laser deposition (PLD) on (0001) Al₂O₃ single crystalline substrates. The values for the oxygen tracer diffusion coefficient were analyzed as a combination of contributions from bulk and interface contributions and compared with results from YSZ/Y₂O₃-multilayers with similar microstructure. Using the Nernst–Einstein equation as the relation between diffusivity and electrical conductivity we find very good agreement between conductivity and diffusion data, and we exclude substantial electronic conductivity in the multilayers. The effect of hetero-interface transport can be well explained by a simple interface strain model. As the multilayer samples consist of columnar film crystallites with a defined inter­face structure and texture, we also discuss the influence of this particular microstructure on the interfacial strain.     

Data Augmentation and Random Multi-Model Deep Learning for Data Classification

In the machine learning (ML) paradigm, data augmentation serves as a regularization approach for creating ML models. The increase in the diversification of training samples increases the generalization capabilities, which enhances the prediction performance of classifiers when tested on unseen examples. Deep learning (DL) models have a lot of parameters, and they frequently overfit. Effectively, to avoid overfitting, data plays a major role to augment the latest improvements in DL. Nevertheless, reliable data collection is a major limiting factor. Frequently, this problem is undertaken by combining augmentation of data, transfer learning, dropout, and methods of normalization in batches. In this paper, we introduce the application of data augmentation in the field of image classification using Random Multi-model Deep Learning (RMDL) which uses the association approaches of multiDL to yield random models for classification. We present a methodology for using Generative Adversarial Networks (GANs) to generate images for data augmenting. Through experiments, we discover that samples generated by GANs when fed into RMDL improve both accuracy and model efficiency. Experimenting across both MNIST and CIAFAR-10 datasets show that, error rate with proposed approach has been decreased with different random models.     

Outlier Detection for Water Supply Data Based on Joint Auto-Encoder

With the development of science and technology, the status of the water environment has received more and more attention. In this paper, we propose a deep learning model, named a Joint Auto-Encoder network, to solve the problem of outlier detection in water supply data. The Joint Auto-Encoder network first expands the size of training data and extracts the useful features from the input data, and then reconstructs the input data effectively into an output. The outliers are detected based on the network’s reconstruction errors, with a larger reconstruction error indicating a higher rate to be an outlier. For water supply data, there are mainly two types of outliers: outliers with large values and those with values closed to zero. We set two separate thresholds, τ₁ and τ₂, for the reconstruction errors to detect the two types of outliers respectively. The data samples with reconstruction errors exceeding the thresholds are voted to be outliers. The two thresholds can be calculated by the classification confusion matrix and the receiver operating characteristic (ROC) curve. We have also performed comparisons between the Joint Auto-Encoder and the vanilla Auto-Encoder in this paper on both the synthesis data set and the MNIST data set. As a result, our model has proved to outperform the vanilla Auto-Encoder and some other outlier detection approaches with the recall rate of 98.94 percent in water supply data.     

Separation of Orbits under Group Actions with an Application to Quantum Systems

Motivated by the task to decide whether two quantum states are equally entangled we consider the orbits under the action of the group of all one-qubit operations. To investigate the orbit structure of this group of local unitary operations we propose to use methods from classical invariant theory as well as new results. Two approaches are presented. The first uses the orbit separation property of invariant rings to distinguish among nonequivalent quantum states. In this context we study the Molien series which describes the structure of the invariant ring as a graded ring. We give a closed formula for the Molien series of the group of one-qubit operations. Our second approach makes use of an equivalence relation, the so-called graph of the action, which relates two points iff they are on the same orbit. For finite groups which factor, are synchronous direct sums, or tensor products we analyze the structure of the graph of the action. This yields new algorithms for the computation of the graph of the action. Received: December 23, 1998; revised version August 27, 1999     

Unsupervised Domain Adaptation Based on Discriminative Subspace Learning for Cross-Project Defect Prediction

Cross-project defect prediction (CPDP) aims to predict the defects on target project by using a prediction model built on source projects. The main problem in CPDP is the huge distribution gap between the source project and the target project, which prevents the prediction model from performing well. Most existing methods overlook the class discrimination of the learned features. Seeking an effective transferable model from the source project to the target project for CPDP is challenging. In this paper, we propose an unsupervised domain adaptation based on the discriminative subspace learning (DSL) approach for CPDP. DSL treats the data from two projects as being from two domains and maps the data into a common feature space. It employs cross-domain alignment with discriminative information from different projects to reduce the distribution difference of the data between different projects and incorporates the class discriminative information. Specifically, DSL first utilizes subspace learning based domain adaptation to reduce the distribution gap of data between different projects. Then, it makes full use of the class label information of the source project and transfers the discrimination ability of the source project to the target project in the common space. Comprehensive experiments on five projects verify that DSL can build an effective prediction model and improve the performance over the related competing methods by at least 7.10% and 11.08% in terms of G-measure and AUC.     

Semi-Supervised Learning with Generative Adversarial Networks on Digital Signal Modulation Classification

Deep Learning (DL) is such a powerful tool that we have seen tremendous success in areas such as Computer Vision, Speech Recognition, and Natural Language Pro-cessing. Since Automated Modulation Classification (AMC) is an important part in Cognitive Radio Networks, we try to explore its potential in solving signal modula-tion recognition problem. It cannot be overlooked that DL model is a complex mod-el, thus making them prone to over-fitting. DL model requires many training data to combat with over-fitting, but adding high quality labels to training data manually is not always cheap and accessible, especially in real-time system, which may counter unprecedented data in dataset. Semi-supervised Learning is a way to exploit unla-beled data effectively to reduce over-fitting in DL. In this paper, we extend Genera-tive Adversarial Networks (GANs) to the semi-supervised learning will show it is a method can be used to create a more data-efficient classifier.     

Analysis and classification of oral tongue squamous cell carcinoma based on Raman spectroscopy and convolutional neural networks

ABSTRACT

To detect oral tongue squamous cell carcinoma (OTSCC) using fibre optic Raman spectroscopy, we present a classification model based on convolutional neural networks (CNN) and support vector machines (SVM). 24 samples Raman spectra of OTSCC and para-carcinoma tissues from 12 patients were collected and analysed. In our proposed model, CNN is used as a feature extractor for forming a representative vector. Then the derived features are fed into an SVM classifier, which is used for OTSCC classification. Experimental results demonstrated that the area under the receiver operating characteristic curve was 99.96% and the classification error was zero (sensitivity: 99.54%, specificity: 99.54%). To show the superiority of this model, comparison results with the state-of-the-art methods showed it can obtain a competitive accuracy. These findings may pay a way to apply the proposed model in the fibre optic Raman instruments for intra-operative evaluation of OTSCC resection margins.     

Graph contractability problem for VLSI layer assignment

Abstract

The layer assignment problem for VLSI routing is the problem of determining which layers can be used for routing the wire segments in the interconnections of nets so that the number of vias is minimized. In this paper, we transform the problem of layer assignment for three‐layer routing to the contractability problem of 3‐colorable graph. This problem is shown to be NP‐complete and a heuristic algorithm is proposed on the basis of the graph contractability model. From our experimental results, the algorithm is faster and efficient to generate very good results. In the average, the number of vias can be reduced by 30 percent or so.     

基于子空间约束的面部特征点跟踪算法

为克服纹理不丰富和非刚性形变等因素引起的面部特征点跟踪困难,提出了一种基于子空间约束的面部特征点跟踪算法。针对人脸运动特点,将面部特征点分为具有复杂运动模式和简单运动模式的特征点集。用通过样例学习得到的特定描述模型准确刻画了具有复杂运动模式的特征点集的变化,保证了子空间约束的有效性。对运动模式简单的特征点集的跟踪则采用基于光流的算法,以提高算法的效率,也为基于特定描述模型的跟踪算法提供了更准确的起始搜索位置。对跟踪结果进一步应用子空间约束解决跟踪中的开孔问题和消除跟踪误差。实验结果表明,在存在较大验部变形和部分特征点纹理不丰富的情况下,该方法可以有效地跟踪较密集的面部特征点。     

The Use of Stress Relaxation Trials to Characterize Tablet Capping

Abstract

A new method is described for estimating capping risks due to the entrapment of air during compression.

For this, the curves obtained during a stress relaxation test were analysed using a WISCHERT model.

Two processing modes of computation from experimental data made it possible to characterize the stress relaxation of each product or mixture of products by a sum of two or three exponentials.

The first exponential represents a more or less rapid structure relaxation. When the quantity of fines particles is too great, the structure relaxation is slowed down by the entrapped air as soon as constant deformation is applied.     

Study of Different Crystalline forms of Mannitol: Comparative Behaviour under Compression

Abstract

Mannitol is an exclpient commonly used in pharmaceutical formulation. Several polymorphic forms have been described but there are only few reports in past literature about their crystalline structures and about their differential properties.

The aim of this work is to prepare and to study the different polymorphic forms. An investigation of commercialized products is then carried out.

The melting points and the water contents are determined and the crystalline structure is differenciated by powder X Ray diffraction.

Four polymorphic forms have been characterized: the a form and the f3 form, which have been obtained in pure state, and the 6 form containing the a form as an impurity, and a fourth form which we have not identified at this time.

Among the commercial products, we have characterized: the g form, the a form and the unidentified form.

The compressibility of these different samples has been studied: the a form is the best. This valuable property has been found again in the commercial α product.

The particle shape seems also to have a great influence upon the compressibility properties. For the same particle size, the granulated powder has a better behaviour than native crystals powders.

Lastly, the possible polymorphic transitions under compression stress during tabletting process has been studied: no transition has been observed.     

Relation extraction with weakly supervised learning based on process-structure-property-performance reciprocity

In this study, we develop a computer-aided material design system to represent and extract knowledge related to material design from natural language texts. A machine learning model is trained on a text corpus weakly labeled by minimal annotated relationship data (~100 labeled relationships) to extract knowledge from scientific articles. The knowledge is represented by relationships between scientific concepts, such as {annealing, grain size, strength}. The extracted relationships are represented as a knowledge graph formatted according to design charts, inspired by the process-structure-property-performance (PSPP) reciprocity. The design chart provides an intuitive effect of processes on properties and prospective processes to achieve the certain desired properties. Our system semantically searches the scientific literature and provides knowledge in the form of a design chart, and we hope it contributes more efficient developments of new materials.     

Progressive image transmission using a prefilter and a modified difference pyramid structure

Abstract

Pyramid data structures have played an important role in progressive image transmission. Over the years, the reduced difference pyramid (RDP) has been found to be one of the best data structure. The RDP takes the differences between the neighboring nodes at the same level. The new modified difference pyramid (MDP) data structure, developed in this paper, takes the differences between successive levels. Simulation results show that both the bit rate and the complexity of the receiver for the MDP structure are lower than those for the RDP structure at the same quality.

An MDP coding process which incorporates a prefilter is also proposed in this paper. Simulation results show it can provide good quality (in a subjective sense) reconstructed images at a lower bit rate than the unfiltered scheme can. Also, acceptable intermediate images are interpolated via the repetition method and via the cubic convolution method to get images which are the same size as the original. Results are compared.     

Distillation separation sequence optimization synthesis based on application of adaptive parallel tabu search

Abstract

Distillation separation sequences can be described as binary tree data structures, because of the analogous structures of distillation separation sequences and binary trees, and then by applying graph theory, the change mechanism of neighborhood separation points based on binary trees is built, correspondingly a kind of highly effective, evolutionary neighborhood structure is constructed. For the purpose of researching further tabu search algorithms, adaptive mechanisms and parallel techniques are introduced. That is, according to memory frequency information, the tabu length and the number of candidates are adaptively adjusted, and multitask parallel technology is realized through the arrangement of the search assignments. The example shows that adaptive parallel tabu searches can solve, successfully, large‐scale distillation separation sequence synthesis problems.     

A DNA Computing Model for the Graph Vertex Coloring Problem Based on a Probe Graph

The biggest bottleneck in DNA computing is exponential explosion, in which the DNA molecules used as data in information processing grow exponentially with an increase of problem size. To overcome this bottleneck and improve the processing speed, we propose a DNA computing model to solve the graph vertex coloring problem. The main points of the model are as follows: ① The exponential explosion problem is solved by dividing subgraphs, reducing the vertex colors without losing the solutions, and ordering the vertices in subgraphs; and ② the bio-operation times are reduced considerably by a designed parallel polymerase chain reaction (PCR) technology that dramatically improves the processing speed. In this article, a 3-colorable graph with 61 vertices is used to illustrate the capability of the DNA computing model. The experiment showed that not only are all the solutions of the graph found, but also more than 99% of false solutions are deleted when the initial solution space is constructed. The powerful computational capability of the model was based on specific reactions among the large number of nanoscale oligonucleotide strands. All these tiny strands are operated by DNA self-assembly and parallel PCR. After thousands of accurate PCR operations, the solutions were found by recognizing, splicing, and assembling. We also prove that the searching capability of this model is up to O(3⁵⁹). By means of an exhaustive search, it would take more than 896 000 years for an electronic computer (5 × 10¹⁴ s⁻¹) to achieve this enormous task. This searching capability is the largest among both the electronic and non-electronic computers that have been developed since the DNA computing model was proposed by Adleman’s research group in 2002 (with a searching capability of O(2²⁰)).