Tech Science Press
Not a member yet
3972 research outputs found
Sort by
Solving Fractional Integro-Differential Equations by Using Sumudu Transform Method and Hermite Spectral Collocation Method
Get PDFIn this paper we are looking forward to finding the approximate analytical solutions for fractional integro-differential equations by using Sumudu transform method and Hermite spectral collocation method. The fractional derivatives are described in the Caputo sense. The applications related to Sumudu transform method and Hermite spectral collocation method have been developed for differential equations to the extent of access to approximate analytical solutions of fractional integro-differential equations
Adversarial Learning for Distant Supervised Relation Extraction
Get PDFRecently, many researchers have concentrated on using neural networks to learn features for Distant Supervised Relation Extraction (DSRE). These approaches generally use a softmax classifier with cross-entropy loss, which inevitably brings the noise of artificial class NA into classification process. To address the shortcoming, the classifier with ranking loss is employed to DSRE. Uniformly randomly selecting a relation or heuristically selecting the highest score among all incorrect relations are two common methods for generating a negative class in the ranking loss function. However, the majority of the generated negative class can be easily discriminated from positive class and will contribute little towards the training. Inspired by Generative Adversarial Networks (GANs), we use a neural network as the negative class generator to assist the training of our desired model, which acts as the discriminator in GANs. Through the alternating optimization of generator and discriminator, the generator is learning to produce more and more discriminable negative classes and the discriminator has to become better as well. This framework is independent of the concrete form of generator and discriminator. In this paper, we use a two layers fully-connected neural network as the generator and the Piecewise Convolutional Neural Networks (PCNNs) as the discriminator. Experiment results show that our proposed GAN-based method is effective and performs better than state-of-the-art methods
Electronic Structure and Physical Characteristics of Dioxin Under External Electric Field
Get PDFDioxin is a highly toxic and caustic substance, which widely existed in the atmosphere, soil and water with tiny particles. Dioxin pollution has become a major problem that concerns the survival of mankind, which must be strictly controlled. The bond length, bond angle, energy, dipole moment, orbital energy level distribution of dioxin under the external field are investigated using DFT (density functional theory) on basis set level of B3LYP/6-31G (d, p). The results indicate that with the increase of the electric field, the length of one Carbon-Oxygen bond increases while another Carbon-Oxygen bond decreases. The energy gradually decreases with the electric field, while the change of the dipole moment has an opposite trend. In the infrared spectra, the vibration frequency decreases with the electric field increasing and shows an obvious red shift. Moreover, the ultraviolet-visible absorption spectra under different electric fields are analyzed with TD-DFT (time-dependent density functional theory) method. The wavelength of the strongest absorption peak increases and occurs red shift with the increase of the electric field. All the above results can provide reference for further research on the properties of dioxin under different external electric field
Semi-Supervised Learning with Generative Adversarial Networks on Digital Signal Modulation Classification
Get PDFDeep 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
An Optimized Labeling Scheme for Reachability Queries
Get PDFAnswering reachability queries is one of the fundamental graph operations. Existing approaches either accelerate index construction by constructing an index that covers only partial reachability relationship, which may result in performing cost traversing operation when answering a query; or accelerate query answering by constructing an index covering the complete reachability relationship, which may be inefficient due to comparing the complete node labels. We propose a novel labeling scheme, which covers the complete reachability relationship, to accelerate reachability queries processing. The idea is to decompose the given directed acyclic graph (DAG) G into two subgraphs, G1 and G2. For G1, we propose to use topological labels consisting of two integers to answer all reachability queries. For G2, we construct 2-hop labels as existing methods do to answer queries that cannot be answered by topological labels. The benefits of our method lie in two aspects. On one hand, our method does not need to perform the cost traversing operation when answering queries. On the other hand, our method can quickly answer most queries in constant time without comparing the whole node labels. We confirm the efficiency of our approaches by extensive experimental studies using 20 real datasets
The Discrete-Analytical Solution Method for Investigation Dynamics of the Sphere with Inhomogeneous Initial Stresses
Get PDFThe paper deals with a development of the discrete-analytical method for the solution of the dynamical problems of a hollow sphere with inhomogeneous initial stresses. The examinations are made with respect to the problem on the natural vibration of the hollow sphere the initial stresses in which is caused by internal and external uniformly distributed pressure. The initial stresses in the sphere are determined within the scope of the exact equations of elastostatics. It is assumed that after appearing this static initial stresses the sphere gets a dynamical excitation and mechanical behavior of the sphere caused by this excitation is described with the so-called three-dimensional linearized equations of elastic wave propagation in initially stressed bodies. For the solution of these equations, which have variable coefficients, the discrete analytical solution method is developed and applied. In particular, it is established that the convergence of the numerical results with respect to the number of discretization is very acceptable and applicable for the considered type dynamical problems. Numerical results on the influence of the initial stresses on the values of the natural frequencies of the hollow sphere are also presented and these results are discussed
Speech Resampling Detection Based on Inconsistency of Band Energy
Get PDFSpeech resampling is a typical tempering behavior, which is often integrated into various speech forgeries, such as splicing, electronic disguising, quality faking and so on. By analyzing the principle of resampling, we found that, compared with natural speech, the inconsistency between the bandwidth of the resampled speech and its sampling ratio will be caused because the interpolation process in resampling is imperfect. Based on our observation, a new resampling detection algorithm based on the inconsistency of band energy is proposed. First, according to the sampling ratio of the suspected speech, a band-pass Butterworth filter is designed to filter out the residual signal. Then, the logarithmic ratio of band energy is calculated by the suspected speech and the filtered speech. Finally, with the logarithmic ratio, the resampled and original speech can be discriminated. The experimental results show that the proposed algorithm can effectively detect the resampling behavior under various conditions and is robust to MP3 compression
Analyzing Cross-domain Transportation Big Data of New York City with Semi-supervised and Active Learning
Get PDFThe majority of big data analytics applied to transportation datasets suffer from being too domain-specific, that is, they draw conclusions for a dataset based on analytics on the same dataset. This makes models trained from one domain (e.g. taxi data) applies badly to a different domain (e.g. Uber data). To achieve accurate analyses on a new domain, substantial amounts of data must be available, which limits practical applications. To remedy this, we propose to use semi-supervised and active learning of big data to accomplish the domain adaptation task: Selectively choosing a small amount of datapoints from a new domain while achieving comparable performances to using all the datapoints. We choose the New York City (NYC) transportation data of taxi and Uber as our dataset, simulating different domains with 90% as the source data domain for training and the remaining 10% as the target data domain for evaluation. We propose semi-supervised and active learning strategies and apply it to the source domain for selecting datapoints. Experimental results show that our adaptation achieves a comparable performance of using all datapoints while using only a fraction of them, substantially reducing the amount of data required. Our approach has two major advantages: It can make accurate analytics and predictions when big datasets are not available, and even if big datasets are available, our approach chooses the most informative datapoints out of the dataset, making the process much more efficient without having to process huge amounts of data
Perfect Quantum Teleportation via Bell States
Get PDFQuantum mechanics shows superiority than classical mechanics in many aspects and quantum entanglement plays an essential role in information processing and some computational tasks such as quantum teleportation (QT). QT was proposed to transmit the unknown states, in which EPR pairs, the entangled states, can be used as quantum channels. In this paper, we present two simple schemes for teleporting a product state of two arbitrary single-particle and an arbitrary two-particle pure entangled state respectively. Alice and Bob have shared an entangle state. Two Bell states are used as quantum channels. Then after Alice measuring her qubits and informing Bob her measurement results, Bob can perfectly reconstruct the original state by performing corresponding unitary operators on his qubits. It shown that a product state of two arbitrary single-particle and an arbitrary two-particle pure entangled state can be teleported perfectly, i.e. the success probabilities of our schemes are both 1
A Fusion Steganographic Algorithm Based on Faster R-CNN
Get PDFThe aim of information hiding is to embed the secret message in a normal cover media such as image, video, voice or text, and then the secret message is transmitted through the transmission of the cover media. The secret message should not be damaged on the process of the cover media. In order to ensure the invisibility of secret message, complex texture objects should be chosen for embedding information. In this paper, an approach which corresponds multiple steganographic algorithms to complex texture objects was presented for hiding secret message. Firstly, complex texture regions are selected based on a kind of objects detection algorithm. Secondly, three different steganographic methods were used to hide secret message into the selected block region. Experimental results show that the approach enhances the security and robustness