Attention mechanism, which selectively focuses on source-side information to learn a context vector for generating target words, has been shown to be an effective method for neural machine translation (NMT). In fact, generating target words depends on not only the source-side information but also the target-side information. Although the vanilla NMT can acquire target-side information implicitly by recurrent neural networks (RNN), RNN cannot adequately capture the global relationship between target-side words. To solve this problem, this paper proposes a novel target-attention approach to capture this information, thus enhancing target word predictions in NMT. Specifically, we propose three variants of target-attention model to directly obtain the global relationship among target words: 1) a forward target-attention model that uses a target attention mechanism to incorporate previous historical target words into the prediction of the current target word; 2) a reverse target-attention model that adopts a reverse RNN model to obtain the entire reverse target words information, and then to combine with source context information to generate target sequence; 3) a bidirectional target-attention model that combines the forward target-attention model and reverse target-attention model together, which can make full use of target words to further improve the performance of NMT. Our methods can be integrated into both RNN based NMT and self-attention based NMT, and help NMT get global target-side information to improve translation performance. Experiments on the NIST Chinese-to-English and the WMT English-to-German translation tasks show that the proposed models achieve significant improvements over state-of-the-art baselines.

Based on the License Assisted Access (LAA) small cell architecture, the LAA coexisting with Wi-Fi heterogeneous networks provide LTE mobile users with high bandwidth efficiency as the unlicensed channels are shared among LAA and Wi-Fi. However, the LAA and Wi-Fi will affect each other when both systems are using the same unlicensed channel in the heterogeneous networks. In such a network, unlicensed band allocation for LAA and Wi-Fi is an important issue that may affect the quality of service (QoS) of both systems significantly. In this paper, we propose an analytical model and conduct simulation experiments to study two allocations for the unlicensed band: unlicensed full allocation (UFA), unlicensed time-division allocation (UTA), and the corresponding buffering mechanism for the LAA data packets. We evaluate the performance for these unlicensed band allocations schemes in terms of the acceptance rate of both LAA and Wi-Fi packet data in LAA buffer queue. Our study provides guidelines for designing channel occupation phase and the buffer size of LAA small cell.

Recently, social relation analysis receives an increasing amount of attention from text to image data. However, social relation analysis from video is an important problem, which is lacking in the current literature. There are still some challenges: 1) it is hard to learn a satisfactory mapping function from low-level pixels to high-level social relation space; 2) how to efficiently select the most relevant information from noisy and unsegmented video. In this paper, we present an Attentive Sequences Recurrent Network model, called ASRN, to deal with the above challenges. First, in order to explore multiple clues, we design a Multiple Feature Attention (MFA) mechanism to fuse multiple visual features (i.e. image, motion, body, and face). Through this manner, we can generate an appropriate mapping function from low-level video pixels to high-level social relation space. Second, we design a sequence recurrent network based on Global and Local Attention (GLA) mechanism. Specially, an attention mechanism is used in GLA to integrate global feature with local sequence feature to select more relevant sequences for the recognition task. Therefore, the GLA module can better deal with noisy and unsegmented video. At last, extensive experiments on the SRIV dataset demonstrate the performance of our ASRN model.

This paper investigates the spectrum allocation problem. Under the current spectrum management mode, large amount of spectrum resource is wasted due to uncertainty of user's demand. To reduce the impact of uncertainty, a presale mechanism is designed based on spectrum pool. In this mechanism, the spectrum manager provides spectrum resource at a favorable price for presale aiming at sharing with user the risk caused by uncertainty of demand. Because of the hierarchical characteristic, we build a spectrum market Stackelberg game, in which the manager acts as leader and user as follower. Then proof of the uniqueness and optimality of Stackelberg Equilibrium is given. Simulation results show the presale mechanism can promote profits for both sides and reduce temporary scheduling.

Key encapsulation mechanism (KEM) combiners, recently formalized by Giacon, Heuer, and Poettering (PKC'18), enable hedging against insecure KEMs or weak parameter choices by combining ingredient KEMs into a single KEM that remains secure assuming just one of the underlying ingredient KEMs is secure. This seems particularly relevant when considering quantum-resistant KEMs which are often based on arguably less well-understood hardness assumptions and parameter choices. We propose a new simple KEM combiner based on a one-time secure message authentication code (MAC) and two-time correlated input secure hash. Instantiating the correlated input secure hash with a t-wise independent hash for an appropriate value of t, yields a KEM combiner based on a strictly weaker additional primitive than the standard model construction of Giaon et al. and furthermore removes the need to do n full passes over the encapsulation, where n is the number of ingredient KEMs, which Giacon et al. highlight as a disadvantage of their scheme. However, unlike Giacon et al., our construction requires the public key of the combined KEM to include a hash key, and furthermore requires a MAC tag to be added to the encapsulation of the combined KEM.

Stance detection in twitter aims at mining user stances expressed in a tweet towards a single or multiple target entities. Detecting and analyzing user stances from massive opinion-oriented twitter posts provide enormous opportunities to journalists, governments, companies, and other organizations. Most of the prior studies have explored the traditional deep learning models, e.g., long short-term memory (LSTM) and gated recurrent unit (GRU) for detecting stance in tweets. However, compared to these traditional approaches, recently proposed densely connected bidirectional LSTM and nested LSTMs architectures effectively address the vanishing-gradient and overfitting problems as well as dealing with long-term dependencies. In this paper, we propose a neural network model that adopts the strengths of these two LSTM variants to learn better long-term dependencies, where each module coupled with an attention mechanism that amplifies the contribution of important elements in the final representation. We also employ a multi-kernel convolution on top of them to extract the higher-level tweet representations. Results of extensive experiments on single and multi-target benchmark stance detection datasets show that our proposed method achieves substantial improvement over the current state-of-the-art deep learning based methods.

In this study, we examined the mechanical durability and self-recovery characterization of liquid crystal gel films with lysine-based gelator. The results indicated that the structural destruction in liquid crystal gel films is attributed to dissociation among network structure. The cracked LC gel films can be recovered by formation of sol-sate films.

The Cuckoo Search (CS) is apt to be trapped in local optimum relating to complex target functions. This drawback has been recognized as the bottleneck of its widespread use. This paper, with the purpose of improving CS, puts forward a Cuckoo Search algorithm featuring Multi-Learning Strategies (LSCS). In LSCS, the Converted Learning Module, which features the Comprehensive Learning Strategy and Optimal Learning Strategy, tries to make a coordinated cooperation between exploration and exploitation, and the switching in this part is decided by the transition probability Pc. When the nest fails to be renewed after m iterations, the Elite Learning Perturbation Module provides extra diversity for the current nest, and it can avoid stagnation. The Boundary Handling Approach adjusted by Gauss map is utilized to reset the location of nest beyond the boundary. The proposed algorithm is evaluated by two different tests: Test Group A(ten simple unimodal and multimodal functions) and Test Group B(the CEC2013 test suite). Experiments results show that LSCS demonstrates significant advantages in terms of convergence speed and optimization capability in solving complex problems.

As a research hotspot and difficulty in the field of computer vision, pedestrian detection has been widely used in intelligent driving and traffic monitoring. The popular detection method at present uses region proposal network (RPN) to generate candidate regions, and then classifies the regions. But the RPN produces many erroneous candidate areas, causing region proposals for false positives to increase. This letter uses improved residual attention network to capture the visual attention map of images, then normalized to get the attention score map. The attention score map is used to guide the RPN network to generate more precise candidate regions containing potential target objects. The region proposals, confidence scores, and features generated by the RPN are used to train a cascaded boosted forest classifier to obtain the final results. The experimental results show that our proposed approach achieves highly competitive results on the Caltech and ETH datasets.

In the obnoxious facility game, we design mechanisms that output a location of an undesirable facility based on the locations of players reported by themselves. The benefit of a player is defined to be the distance between her location and the facility. A player may try to manipulate the output of the mechanism by strategically misreporting her location. We wish to design a λ-group strategy-proof mechanism i.e., for every group of players, at least one player in the group cannot gain strictly more than λ times her primary benefit by having the entire group change their reports simultaneously. In this paper, we design a k-candidate λ-group strategy-proof mechanism for the obnoxious facility game in the metric defined by k half lines with a common endpoint such that each candidate is a point in each of the half-lines at the same distance to the common endpoint as other candidates. Then, we show that the benefit ratio of the mechanism is at most 1+2/(k-1)λ. Finally, we prove that the bound is nearly tight.

Despite the widespread use of deep learning for speech emotion recognition, they are severely restricted due to the information loss in the high layer of deep neural networks, as well as the degradation problem. In order to efficiently utilize information and solve degradation, attention-based dense long short-term memory (LSTM) is proposed for speech emotion recognition. LSTM networks with the ability to process time series such as speech are constructed into which attention-based dense connections are introduced. That means the weight coefficients are added to skip-connections of each layer to distinguish the difference of the emotional information between layers and avoid the interference of redundant information from the bottom layer to the effective information from the top layer. The experiments demonstrate that proposed method improves the recognition performance by 12% and 7% on eNTERFACE and IEMOCAP corpus respectively.

The electric properties of a field-effect transistor not only depend on gate surface sidewall but also on channel orientation when applying channel stain engineering. The change of the gate surface and channel orientation through the rotated FinFETs provides the capability to compare the orientation dependence of performance and reliability. This study characterized the <100> and <110> channels of FinFETs on the same wafer under tensile and compressive stresses by cutting the wafer into rectangular silicon pieces and evaluated their piezoresistance coefficients. The piezoresistance coefficients of the <100> and <110> silicon under tensile and compressive stresses were first evaluated based on the current setup. Tensile stresses enhance the mobilities of both <100> and <110> channels, whereas compressive stresses degrade them. Electrical characterization revealed that the threshold voltage variation and drive current degradation of the {100} surface were significantly higher than those of {110} for positive bias temperature instability and hot carrier injection with equal gate and drain voltage (VG=VD). By contrast, insignificant difference is noted for the subthreshold slope degradation. These findings imply that a higher ratio of bulk defect trapping is generated by gate voltage on the <100> surface than that on the <110> surface.

This study proposes a novel machine learning architecture and various learning algorithms to build-in anti-phishing services for avoiding cyber-phishing attack. For the rapid develop of information technology, hackers engage in cyber-phishing attack to steal important personal information, which draws information security concerns. The prevention of phishing website involves in various aspect, for example, user training, public awareness, fraudulent phishing, etc. However, recent phishing research has mainly focused on preventing fraudulent phishing and relied on manual identification that is inefficient for real-time detection systems. In this study, we used methods such as ANOVA, X2, and information gain to evaluate features. Then, we filtered out the unrelated features and obtained the top 28 most related features as the features to use for the training and evaluation of traditional machine learning algorithms, such as Support Vector Machine (SVM) with linear or rbf kernels, Logistic Regression (LR), Decision tree, and K-Nearest Neighbor (KNN). This research also evaluated the above algorithms with the ensemble learning concept by combining multiple classifiers, such as Adaboost, bagging, and voting. Finally, the eXtreme Gradient Boosting (XGBoost) model exhibited the best performance of 99.2%, among the algorithms considered in this study.

In this paper, we propose a method for enhancing the visibility of omnidirectional spherical images by enlarging the foreground and compressing the background without provoking a sense of visual incompatibility by using a simplified spring model.

We analyze the behaviors of the FXLMS algorithm using a statistical-mechanical method. The cross-correlation between a primary path and an adaptive filter and the autocorrelation of the adaptive filter are treated as macroscopic variables. We obtain simultaneous differential equations that describe the dynamical behaviors of the macroscopic variables under the condition that the tapped-delay line is sufficiently long. The obtained equations are deterministic and closed-form. We analytically solve the equations to obtain the correlations and finally compute the mean-square error. The obtained theory can quantitatively predict the behaviors of computer simulations including the cases of both not only white but also nonwhite reference signals. The theory also gives the upper limit of the step size in the FXLMS algorithm.

The-state-of-the-art neural quality estimation (QE) of machine translation model consists of two sub-networks that are tuned separately, a bidirectional recurrent neural network (RNN) encoder-decoder trained for neural machine translation, called the predictor, and an RNN trained for sentence-level QE tasks, called the estimator. We propose to combine the two sub-networks into a whole neural network, called the unified neural network. When training, the bidirectional RNN encoder-decoder are initialized and pre-trained with the bilingual parallel corpus, and then, the networks are trained jointly to minimize the mean absolute error over the QE training samples. Compared with the predictor and estimator approach, the use of a unified neural network helps to train the parameters of the neural networks that are more suitable for the QE task. Experimental results on the benchmark data set of the WMT17 sentence-level QE shared task show that the proposed unified neural network approach consistently outperforms the predictor and estimator approach and significantly outperforms the other baseline QE approaches.

In this letter, we explored the usage of spatio-temporal information in one unified framework to improve the performance of multichannel speech recognition. Generalized cross correlation (GCC) is served as spatial feature compensation, and an attention mechanism across time is embedded within long short-term memory (LSTM) neural networks. Experiments on the AMI meeting corpus show that the proposed method provides a 8.2% relative improvement in word error rate (WER) over the model trained directly on the concatenation of multiple microphone outputs.

Sponsored search is a mechanism that shows the appropriate advertisements (ads) according to search queries. The orders and payments of ads are determined by the auction. However, the externalities which give effects to CTR and haven't been considered in some existing works because the mechanism with externalities has high computational cost. In addition, some algorithms which can calculate the approximated solution considering the externalities within the polynomial-time are proposed, however, it assumed that one bidder can propose only a single ad. In this paper, we propose the approximation allocation algorithm that one bidder can offer many ads considering externalities. The proposed algorithm employs the concept of the combinatorial auction in order to consider the combinational bids. In addition, the proposed algorithm can find the approximated allocation by the dynamic programming. Moreover, we prove the computational complexity and the monotonicity of the proposed mechanism, and demonstrate computational costs and efficiency ratios by changing the number of ads, slots and maximum bids. The experimental results show that the proposed algorithm can calculate 0.7-approximation solution even though the full search can't find solutions in the limited times.

Wireless engineers and business planners commonly raise the question on where, when, and how millimeter-wave (mmWave) will be used in 5G and beyond. Since the next generation network is not just a new radio access standard, but also an integration of networks for vertical markets with diverse applications, answers to the question depend on scenarios and use cases to be deployed. This paper gives four 5G mmWave deployment examples and describes in chronological order the scenarios and use cases of their probable deployment, including expected system architectures and hardware prototypes. The first example is a 28 GHz outdoor backhauling for fixed wireless access and moving hotspots, which will be demonstrated at the PyeongChang Winter Olympic Games in 2018. The second deployment example is a 60 GHz unlicensed indoor access system at the Tokyo-Narita airport, which is combined with Mobile Edge Computing (MEC) to enable ultra-high speed content download with low latency. The third example is mmWave mesh network to be used as a micro Radio Access Network (µ-RAN), for cost-effective backhauling of small-cell Base Stations (BSs) in dense urban scenarios. The last example is mmWave based Vehicular-to-Vehicular (V2V) and Vehicular-to-Everything (V2X) communications system, which enables automated driving by exchanging High Definition (HD) dynamic map information between cars and Roadside Units (RSUs). For 5G and beyond, mmWave and MEC will play important roles for a diverse set of applications that require both ultra-high data rate and low latency communications.

Authors previously studied the degradation of electrical contacts under the condition of various external micro-oscillations. They also developed a micro-sliding mechanism (MSM2), which causes micro-sliding and is driven by a piezoelectric actuator and elastic hinges. Using the mechanism, experimental results were obtained on the minimal sliding amplitude (MSA) required to make the electrical resistance fluctuate under various conditions. In this paper, to develop a more realistic model of input waveform than the previous one, Ts/2 is set as the rising or falling time, Tc as the flat time, and τ/2 as the duration in a sliding period T (0.25 s) of the input waveform. Using the Duhamel's integral method and an optimization method, the physical parameters of natural angular frequency ω0 (12000 s-1), damping ratio ζ (0.05), and rising and falling time Ts (1.3 or 1.2 ms) are obtained. Using the parameters and the MSA, the total acceleration of the input TA (=f(t)) and the displacement of the output x(t) are also obtained using the Fourier series expansion method. The waveforms x(t) and the experimental results are similar to each other. If the effective mass m, which is defined as that of the movable parts in the MSM2, is 0.1 kg, each total force TF (=2mTA) is estimated from TA and m. By the TF, the cases for 0.3 N/pin as frictional force or in impulsive as input waveform are more serious than the others. It is essential for the safety and the confidence of electrical contacts to evaluate the input waveform and the frictional force. The ringing waveforms of the output displacements x(t) are calculated at smaller values of Ts (1.0, 0.5, and 0.0 ms) than the above values (1.3 or 1.2 ms). When Ts is slightly changed from 1.3 or 1.2 ms to 1.0 ms, the ringing amplitude is doubled. For the degradation of electrical contacts, it is essential that Ts is reduced in a rectangular and impulsive input. Finally, a very simple wear model comprising three stages (I, II, and III) is introduced in this paper. Because Ts is much shorter in a rectangular or impulsive input than in a sinusoidal input, it is considered that the former more easily causes wear than the latter owing to a larger frictional force. Taking the adhesive wear in Stages I and III into consideration, the wear is expected to be more severe in the case of small damped oscillations owing to the ringing phenomenon.