Search

Most existing AMBTC (absolute moment block truncation coding)-based reversible data hiding (RDH) schemes cannot be decoded by traditional AMBTC decoders, because they change the coding structure of the AMBTC compressed code stream, which is unidentified for decoders. Even if some AMBTC-based RDH methods can be decoded by traditional AMBTC decoders, but the obtained payload is low. To this end, in this paper, a high-capacity and decodable AMBTC-based RDH scheme is presented. It is well observed that an $m\times n$ -sized bitmap has $2^{m\times n}$ combinations of ‘1’ and ‘0’. However, all the bitmaps of an image occupy only a small part of all the combinations, and a large part are not used in this image. Motivated by this observation, we propose a one-to-many map between the used and unused combinations, in which one used bitmap can be mapped to multiple unused combinations, achieving high payloads (e.g., $\log _{2} 25$ bits for one bitmap of Lena), and more importantly, completely reserving the coding structure of the AMBTC-compressed code. Unlike existing decodable AMBTC-based RDH methods only capable of achieving a fixed payload for one test image, our method can adjust adaptively payloads according to the required visual quality by introducing a predefined threshold. The experimental results also demonstrate that our proposed scheme can largely increase the payload on the basis of maintaining good visual quality.

Digital microfluidic biochips (DMFBs) are increasingly important and are used for point-of-care, drug discovery, clinical diagnosis, immunoassays, etc. Pin-constrained DMFBs are an important part of digital microfluidic biochips, and they have gained increasing attention from researchers. However, many previous works have focused on the problem of electrode addressing and aimed to minimize the number of control pins in pin-constrained DMFBs. Although the number of control pins can be effectively redistributed through broadcast addressing technology, the chip reliability will be reduced if the signals are shared arbitrarily. Arbitrary signal sharing can lead to a large number of actuations for many idle electrodes, and as a result, a trapping charge or decreasing contact angle problem could occur for some electrodes, reducing the reliability of the chip. To address this problem, the appropriate electrode matching object should be carefully selected, and the influence of these factors on chip reliability should be fully considered. For this purpose, we aimed to fully consider electrode addressing and the reliability of the chip in improving the reliability of DMFBs. This paper proposed a pin addressing method based on a support vector machine (SVM) with the reliability constraint algorithm, which can fully consider the electrode addressing method and the reliability of the chip together. The proposed method achieved an average maximum number of electrode actuations that was 53.8% and 18.2% smaller than those of the baseline algorithm and the graph-based algorithm, respectively. The simulation experiment results showed that the proposed method can efficiently solve reliability problems during the DMFB design process.

The central support tilting thrust bearing is widely used in the motor-generator unit, and the central support tilting thrust bearing performs poorly compared with the offset bearing. In order to improve the performance of the thrust bearing capable of the bidirectional operation, a spring-supported switching tilting thrust bearing was designed, which could run under offset condition at both clockwise and counterclockwise. In order to verify the reliability of the bearing, we designed a true size thrust bearing test bench and measured the performance parameters of the temperature, pressure, oil film thickness, and power loss of the pad. This article introduces in detail the operation mechanism of the bidirectional offset spring-bed tilting thrust bearing. The bidirectional offset spring-bed tilting thrust bearing structure and the central support bearing structure were compared and tested. The test results of the performance difference of large tilting pad bearings with different structures were obtained. According to the experiment, the spring support structure has good adaptability, and the improved bidirectional offset support bearing not only has higher bearing capacity, but also has better performance in all aspects of temperature rise and loss than the central support bearing. The actual size experiment provides the experimental data for the theoretical calculation of large tilting pad bearings, providing a more accurate basis for the bearing performance and safety assessment.

Text sentiment classification is an essential research field of natural language processing. Recently, numerous deep learning-based methods for sentiment classification have been proposed and achieved better performances compared with conventional machine learning methods. However, most of the proposed methods ignore the interactive relationship between contextual semantics and sentimental tendency while modeling their text representation. In this paper, we propose a novel Interactive Dual Attention Network (IDAN) model that aims to interactively learn the representation between contextual semantics and sentimental tendency information. Firstly, we design an algorithm that utilizes linguistic resources to obtain sentimental tendency information from text and then extract word embeddings from the BERT (Bidirectional Encoder Representations from Transformers) pretraining model as the embedding layer of IDAN. Next, we use two Bidirectional LSTM (BiLSTM) networks to learn the long-range dependencies of contextual semantics and sentimental tendency information, respectively. Finally, two types of attention mechanisms are implemented in IDAN. One is multihead attention, which is the next layer of BiLSTM and is used to learn the interactive relationship between contextual semantics and sentimental tendency information. The other is global attention that aims to make the model focus on the important parts of the sequence and generate the final representation for classification. These two attention mechanisms enable IDAN to interactively learn the relationship between semantics and sentimental tendency information and improve the classification performance. A large number of experiments on four benchmark datasets show that our IDAN model is superior to competitive methods. Moreover, both the result analysis and the attention weight visualization further demonstrate the effectiveness of our proposed method.

Aspect-based sentiment classification is a fine-grained task in sentiment analysis, it aims to detect the sentiment polarity of an aspect in a given context. Existing approaches mostly focused on the processing of aspect words, which often fail to adequately model the context via aspect terms. However, the surrounding words in context usually have an important impact on the sentiment polarity of aspect words. This paper proposes a context enhanced attention network (CE-ATT) to enhance context modeling for aspect-based sentiment classification. Firstly, the text sequence was divided into left and right contexts by a given aspect. Then, the Bidirectional Long Short Term Memory (BiLSTM) was used to model the left context and aspect, the right context and aspect, and both sides context and aspect, respectively. After summing these contextual representations, an attention mechanism was used to concentrate on the parts that have an important influence on the aspect word. The experimental results on three benchmark datasets demonstrate the effectiveness of the CE-ATT method.

The traditional calibration method for charging piles which used for electric vehicles relies on the field inspection of standard devices of metrology agency. However, with the continuous expansion of electric vehicles' scale in the future, the calibration workload of charging piles will become larger and larger, calibration method that relies solely on the standard devices cannot hold so heavy workload. In this paper, we propose a new calibration method, which can reduce the calibration workload of standard devices fundamentally. Firstly, we train a model which called virtual standard device for the metrology standard through machine learning, and when calibrating the charging pile, we just need to input the charging message into the model, and then we can get the calibration result which includes normal pile, abnormal pile and uncertain pile. Once the model training is completed, the entire calibrating process including the collection of message data, the import of data into the model and the calibration result requires no human intervention unless the result is uncertain. Moreover, it is found that the higher the model accuracy is, the smaller the uncertainty interval is and the less the manual calibration work is. The method of simulated calibration plus field calibration proposed in this paper can help reduce the calibration workload fundamentally and greatly reduce the calibrating pressure of metrology agency.

Digital micro-fluidic biochips (DMFBs) are revolutionizing laboratory procedures for point-of-care clinical diagnostics, environmental monitoring, and protein analysis. Those procedures require high precision of the output for every operation so it need to ensure the chip reliability and the chip lifetime. Because the electrodes on the chip may be reused at different experimental stages, a degraded electrode may be reused for many times. So the lifetime of the chips is closely related to the total actuating time of an electrode. Thus, the electrode total actuating time needs to be considered carefully in an efficient DMFBs design process. This paper proposed an improved whale optimization algorithm (IWOA), which can reduce excessive use of an electrode and reuse electrodes in average manner to optimal the longest lifetime of DMFBs during the design process. Firstly, the position mass of individual whales was improved to solve the problem that the data, calculated by the WOA, cannot be directly used to represent the sequence of operations. Secondly, the inertial weight was added to enhance the local search ability of WOA. The simulation experimental results showed that this algorithm was able to solve lifetime optimization problems. The maximum electrode time used was reduced by 10% to 20%. The performance of efficiency and convergence of algorithm were very good.

Aiming at the non-optimal performance of SFD under different operating conditions, this paper proposes a controllable clearance squeeze film damper (CCSFD) that uses hydraulic pressure to adjust the radial clearance and establishes a corresponding test rig. Measure the critical amplitude under different imbalances and hydraulic pressures. Give the calculation formula of hydrodynamic force by directly integrating pressure distribution and build the corresponding numerical model. Both experimental and numerical results show that CCSFD can achieve synchronous adjustment of the damper's performance under a non-stop state of the rotor system. In this case, the radial clearance can transform actively in the range of 1.75‰-3.19‰, which carries out a noticeable improvement of the damper's performance and achieves satisfactory performance under different operating conditions.

Gene function prediction is used to assign biological or biochemical functions to genes, which continues to be a challenging problem in modern biology. Genes may exhibit many functions simultaneously, and these functions are organized into a hierarchy, such as a directed acyclic graph (DAG) for Gene Ontology (GO). Because of these characteristics, gene function prediction can be seen as a typical hierarchical multi-label classification (HMC) task. A novel HMC method based on neural networks is proposed in this article for predicting gene function based on GO. The proposed method belongs to a local approach by transferring the HMC task to a set of subtasks. There are three strategies implemented in this method to improve its performance. First, to tackle the imbalanced data set problem when building the training data set for each class, negative instances selecting policy and SMOTE approach are used to preprocess each imbalanced training data set. Second, a particular multi-layer perceptron (MLP) is designed for each node in GO. Third, a post processing method based on the Bayesian network is used to guarantee that the results are consistent with the hierarchy constraint. The experimental results indicate that the proposed HMC-MLPN method is a promising method for gene function prediction based on a comparison with two other state-of-the-art methods.

The state of health (SOH) is a key parameter for fault diagnoses and safety early warnings in the life cycle of lithium batteries in electric vehicles. The SOH prediction model generally uses the experimental data from the same batch of batteries in the same environment. These data may cause “overfitting” to the model as the attenuation of lithium batteries varies depending on the batch and working condition, especially in actual use. And there is a risk of serious deviation in the prediction result if there is no true value of the model. This paper proposes a SOH prediction model that evaluates the prediction uncertainty using data from different batches of batteries under actual working conditions. It not only quantitatively evaluates the credibility of the prediction model in absence of true values, but also filtering training data to improve the model accuracy and avoid overfitting. The model produces evaluation uncertainty for the prediction result based on the Gaussian process regression (GPR) method. Experiments' results show that the evaluation uncertainty is better than the prediction variance of GPR. The accuracy of the prediction model using the minimum evaluation uncertainty as the training data screening is an order of magnitude higher than that using all data for training.

For the open laboratory management of an university, the attendance data is not accurate, and meanwhile, the statistics data is too easy to be lacked by using the traditional devices. Wi-Fi probe for people flow detection is often used for safety warnings in public places and commercial promotion of products, laboratory attendance and so on. However, the existing technologies used for laboratory attendance only identify the presence of a target while neglect the fact that when the coverage range of Wi-Fi signal exceeds the attendance range, the attendance data obtained by Wi-Fi probe will be inaccurate. So far, Wi-Fi probe has not been applied for managing people flow of open laboratory. In this paper, we propose an algorithm based on Wi-Fi probe so as to successfully increase the quality of people flow detection in open laboratory. We use RSSI positioning theory to distinguish people inside and outside the laboratory and identify the statistics of the people flow. The experimental results also demonstrate our system not only has higher accuracy than the existing RFID attendance systems, but also provide the rank of time which people spent in the laboratory, the numbers of people and the total time which all the people spend in the lab in a certain time. Managers and users can log in the system to check the statistical data. The system improves the utilization of the laboratory and contributes to the development of open laboratory management practices.

This paper presents an enhanced sparse based discriminative topic representation method for text categorization. By constructing category center vectors and combining with the latent Dirichlet allocation model, a more discriminative dictionary is obtained, which can describe the relationship between topic and word well. Furthermore, an enhanced sparse representation of documents can be generated with a $L_{1/2}$ regularization in order to achive a good relationship between document and topic. The experimental results show that our proposed approach achieves more stable classification performance and obtains more high sparse degree.

Collaborative deep learning can provide high learning accuracy even participanted users' datasets are small. In the training process, users only share their locally obtained parameters, therefore it is believed that the privacy of users' original datasets can be protected. However, we present an attack approach against users' privacy in collaborative deep learning by utilizing Generative Adversarial Network (GAN) and Membership Inference. In this attack, an attacker builds a discriminator based on users' shared parameters and then trains a GAN network locally. The GAN can refactor the training records of the collaborative deep learning system. According to the generated records, the attacker uses the extent of model overfitting on an input and gets the membership of each group of records by the simplified Membership Inference attack. We evaluate the presented attack model over datasets of complex representations of handwritten digits (MINIST) and face images (CelebA). The results show that an attacker can easily generate the original training sets and classify them to obtain the membership between users' records and their identities in the collaborative deep learning.

Digital microfluidic biochips (DMFBs) are attractive instruments for obtaining modern molecular biology and chemical measurements. Due to the increasingly complex measurements carried out on a DMFB, such chips are more prone to failure. To compensate for the shortcomings of the module-based DMFB, this paper proposes a routing-based fault repair method. The routing-based synthesis methodology ensures a much higher chip utilization factor by removing the virtual modules on the chip, as well as removing the extra electrodes needed as guard cells. In this paper, the routing problem is identified as a dynamic path-planning problem and mixed path design problem under certain constraints, and an improved Dijkstra and improved particle swarm optimization (ID-IPSO) algorithm is proposed. By introducing a cost function into the Dijkstra algorithm, the path-planning problem under dynamic obstacles is solved, and the problem of mixed path design is solved by redefining the position and velocity vectors of the particle swarm optimization. The ID-IPSO routing-based fault repair method is applied to a multibody fluid detection experiment. The proposed design method has a stronger optimization ability than the greedy algorithm. The algorithm is applied to , , and fault-free chips. The proposed ID-IPSO routing-based chip design method saves 13.9%, 14.3%, and 14.5% of the experiment completion time compared with the greedy algorithm. Compared with a modular fault repair method based on the genetic algorithm, the ID-IPSO routing-based fault repair method has greater advantages and can save 39.3% of the completion time on average in the completion of complex experiments. When the ratio of faulty electrodes is less than 12% and 23%, the modular and ID-IPSO routing-based fault repair methods, respectively, can guarantee a 100% failure repair rate. The utilization rate of the electrodes is 18% higher than that of the modular method, and the average electrode usage time is 17%. Therefore, the ID-IPSO routing-based fault repair method can accommodate more faulty electrodes for the same fault repair rate, the experiment completion time is shorter, the average number of electrodes is lower, and the security performance is better.

The available capacity of a lithium battery reflects its actual capacity under certain constraints. It serves as an important deciding factor for the electric vehicles’ energy management system. Online estimation allows the construction of a mathematical model with easily measurable variables as input to estimate the main variables that are difficult to measure in a complex environment. Current research applies the trained model of another battery to the current testing battery. Due to differences across batteries, the same model is unable to deliver an accurate description of the chemical processes of the same type of batteries. In order to solve the problem of model adaptability, an iterative EKF-GPR method is proposed based on the full charging curve of the battery being estimated and the daily charging voltage, current and time data based on the assumption that there are no abrupt changes in the lithium battery charging curves. The method is to establish a model for the estimation of the available current charging capacity of the lithium battery. It converts macro-time machine learning into the prediction and estimation of single charging curve on the micro-time scale, thus guaranteeing the accuracy of available capacity estimation models for any batch of batteries.

This paper presents a novel portable chemical oxygen demand (COD) detection method and instrument design based on the microfluidic chip technology. The oxidation reaction and temperature detection are integrated in a microfluidic chip. The principle is that there is heat released during the oxidation of organic compounds and the heat is proportional to the amount of organic compounds contained in water sample. The heat variation can be converted to voltage signal which has linear relationship with COD value. In this research, the NaClO solution is employed as oxidizer and the glucose solution is served as analyte. Three different geometric microfluidic chips are designed in order to choose the most suitable microfluidic chip scheme, the oxidizer concentration and injection rates. The flows characteristics, mixing behavior and heat release of fluids within the microfluidic chip are simulated and investigated by COMSOL multi-physics simulation software. The results show that the most suitable NaClO concentration, injection rates and microfluidic chip scheme are 28mol/L, lmm/s and Y sharp with block. A portable COD detection instrument based on the microfluidic chip technology is developed. There is a simple experiment for proving the feasibility of the instrument, the experiment results show that the experiment device is work well.

In text categorization, feature representation for dimensionality reduction is a key step. Usually, some commonly used methods, e.g., latent semantic analysis (LSA), yield a dense representation or a dense transformation matrix, which is difficult to precisely characterize the document-topic or the topic-word relationship. This paper proposes a novel discriminative topic sparse representation (DTSR) approach for text categorization, in which two stages are included: the topic dictionary construction and sparse representation. Firstly, a discriminative and interpretable dictionary is constructed to characterize the topic-word relationship. The dictionary contains all category center vectors as well as some semantic topic vectors generated by a latent Dirichlet allocation (LDA) model. Furthermore, each document can be represented with a sparse form to obtain a good document-topic relationship. Experimental results on well-known benchmark datasets indicate that the proposed method not only achieves a satisfactory classification performance but also provides a reasonable sparse semantic meaningful.

The localized surface plasmons of metallic nanoparticles are able to concentrate light into small volumes, which lead to a variety of fundamental studies and practical applications in plasmonics. For example, by strong coupling between metallic nanoparticles, plasmonic antennas are able to concentrate and re-emit light in a controllable way. A variety of structures of optical antennas have been investigated in in the past decade. The near- and far-field responses of the plasmonic nanoantennas for example, the intensity and phase distributions, and the emission polarization state are found to be sensitive to polarization. This sensitivity is determined to arise from structural properties including particle size, shape, spacing, relative positions and symmetry of nanoparticles. In this review, we will discuss our recent advances in plasmonic nanoparticle antennas from the polarization point of view, i.e., control of the incident polarization-dependent near-field enhancement, control of the (far-field) polarization of elastic or inelastic scattering light, and outlook the corresponding impacts in understanding physics and nanophotonic devices applications.

The grouped structure has successfully been embedded in sparse models for feature selection; however, some groups generated by clustering method might be difficult to interpret their semantic information if the number of words in the group is very large. This paper proposes a novel approach in which a group structure is constructed and its corresponding sparse model is used to select features for text categorization. After variable preselection, an algorithm is developed to generate groups, such that each group only contains two or three closely related words, which can reflect more essential semantic meaningful. Finally, structural sparse mode is used to select feature in wrapper way. The experimental results demonstrate that the proposed method achieves comparable precision and improves the sparsity considerably, which means that the model has better interpretability.

Digital microfluidic biochip (DMFB) is an attractive platform for immunoassays, point-of-care clinical diagnostics, DNA sequencing, and other laboratory procedures in laboratory experiments due to its flexible application and low fabrication cost and further for the development of instruments. As being applied for these safe-critical applications, on-line testing methods are required to ensure robust DMFB operation and high confidence in the outcome of experiments. Prior work on DMFB testing paid little attention to on-line testing which test droplet will also work during bioassay, but it's necessary and important because some defects may manifest during field operation. So we proposed a new heuristic way by improved ant colony algorithm (IACA) to get the optimal routing path under fluidic constraints. Both static and dynamic fluidic constraints are considered in our testing method. The simulation results showed our method is convergent and could be applied in various operations to reduce about 5% testing time.

Gene function prediction is a complicated and challenging hierarchical multi-label classification (HMC) task, in which genes may have many functions at the same time and these functions are organized in a hierarchy. This paper proposed a novel HMC algorithm for solving this problem based on the Gene Ontology (GO), the hierarchy of which is a directed acyclic graph (DAG) and is more difficult to tackle. In the proposed algorithm, the HMC task is firstly changed into a set of binary classification tasks. Then, two measures are implemented in the algorithm to enhance the HMC performance by considering the hierarchy structure during the learning procedures. Firstly, negative instances selecting policy associated with the SMOTE approach are proposed to alleviate the imbalanced data set problem. Secondly, a nodes interaction method is introduced to combine the results of binary classifiers. It can guarantee that the predictions are consistent with the hierarchy constraint. The experiments on eight benchmark yeast data sets annotated by the Gene Ontology show the promising performance of the proposed algorithm compared with other state-of-the-art algorithms.

Hyperspectral image (HSI) classification draws a lot of attentions in the past decades. The classical problem of HSI classification mainly focuses on a single HSI scene. In recent years, cross-scene classification becomes a new problem, which deals with the classification models that can be applied across different but highly related HSI scenes sharing common land cover classes. This paper presents a cross-scene classification framework combining spectral–spatial feature extraction and manifold-constrained feature subspace learning. In this framework, spectral–spatial feature extraction is completed using three-dimensional (3D) wavelet transform while manifold-constrained feature subspace learning is implemented via multitask nonnegative matrix factorization (MTNMF) with manifold regularization. In 3D wavelet transform, we drop some coefficients corresponding to high frequency in order to avoid data noise. In feature subspace learning, a common dictionary (basis) matrix is shared by different scenes during the nonnegative matrix factorization, indicating that the highly related scenes should share than same low-dimensional feature subspace. Furthermore, manifold regularization is applied to force the consistency across the scenes, i.e. all pixels representing the same land cover class should be similar in the low-dimensional feature subspace, though they may be drawn from different scenes. The experimental results show that the proposed method performs well in cross-scene HSI datasets.

This paper proposes a novel spectral variable selection approach named similar offspring voting genetic algorithm (sovGA). Unlike the traditional voting method (here called voting genetic algorithm) that attempts to overcome the stochastic risk of a genetic algorithm but may damage its optimal variable combination, sovGA aims to achieve a trade-off between prediction precision and reliability of models. Genetic algorithm was repeated many times, and some effective similar offspring were obtained by a proposed algorithm; then, the spectral variables were selected by a voting mechanism. Two calibration methods, partial least square regression and multiple linear regression, were constructed to evaluate performance. The experimental results show that sovGA outperformed genetic algorithm, voting genetic algorithm, and the full spectrum situation in most cases.

In order to research the effect of different layout forms of steel plate on the axial compression behavior of a steel plate-concrete composite shear wall, this paper presents the experimental results and analysis of the axial compression behavior of a composite shear wall, with different layout forms of steel plate. A total of three tests were carried out, composed of two composite walls with built-in steel plate, and one composite wall with two skins of steel plate. The gross dimensions of the three specimens were 1206 mm × 2006 mm × 300 mm. Experimental results show that the composite wall with two skins of steel plate has an optimal ability of elastic-plastic deformation, and the maximum axial compressive bearing capacity among the three specimens. Using the energy method, the critical local buckling stresses of steel plate were calculated, and compared with the yield stresses. According to different confined actions of concrete, concrete constitutive models were proposed, and the axial compressive strengths of confined concrete were calculated. Considering the local buckling of steel plate and confined concrete, the calculation formula of the axial compression of the composite wall was put forward, and the calculated results were in good agreement with the test results. Therefore, the different layout forms of steel plate have a great influence on its buckling, and on the concrete inhibition effect, which can affect the axial compressive bearing capacity of the composite wall.