A script, which describes the evolutionary path of events, is a structured event sequence. Script event prediction aims to predict the next event from a sequence of historical events. Current studies favor modeling macroscale information, including event sequence, event unit, and event argument, while ignoring the smallest unit of event, i.e., argument vocabulary, which we refer to as microscale information. To fuse event information from different scales, we propose a staged encoder-decoder architecture (SEDA) for script event prediction. SEDA aggregates microscale information to enhance the representation of event arguments and event units in the event-enhancement stage, and extracts the context of event sequence to predict the most relevant candidate event in the context-extraction stage. Both stages of SEDA adopt an efficient and scalable encoder-decoder architecture. The experimental results demonstrate that the accuracy of SEDA on the MCNC task surpasses that of the current SOTA baseline. Additionally, we empoly a method based on the Shapley value to calculate the importance of different event units in an event sequence, providing a quantitative analysis of the prediction results.

Multi-agent reinforcement learning (MARL) algorithms with sequential decision-making strategies have achieved great success in cooperation tasks recently. To overcome the non-stationarity problem, these methods design a centralized controller that takes global observation as input and chooses actions for each agent in sequence. However, in most scenarios, global information is only available at training time, while agents act synchronously with their local observation at execution time, which prevents agents from leveraging more information in cooperation. In this paper, based on actor-critic architecture, we propose the actor-predicts-critic (APC) algorithm, in which the actor learns to predict the global representations of centralized critic from local observation. During the training, the actor not only receives the estimated state values, but also takes the critic’s representations that are extracted from global information as the prediction targets. Since these global representations are closely related to agents’ goals and rewards, agents can achieve better cooperation on MARL tasks utilizing the predicted representations. To prove the validity of APC, we evaluate the algorithm on StarCraft2, Google Research Football, and MultiAgent Mujoco benchmarks. The results show that APC significantly outperforms the strong baselines in centralized training and decentralized execution (CTDE) framework, including MATDec, MAPPO, and fine-tuned QMIX.

This paper proposes a novel contact force sensing and control method for the inserting operation during precise assembly process, which is based on a micromanipulator integrated with force sensors. At first, theoretical analysis is carried out to calculate the admissible contact force between the gripped holes and the pegs. The contact force thresholds which are smaller than the admissible contact forces are adopted in the control algorithm to avoid the rotating of the gripped holes during assembly process. The force sensors are calibrated using an ATI force sensor and the conversing coefficients are calculated. The admissible contact forces are tested when different contact distance and preload force are adopted. The performance of the proposed contact force sensing and control method is verified by carrying out the task of applying contact force on the surface of the gripped holes with different contacting speeds. The results indicate that the contact force can be adjusted to be smaller than the threshold 1 and the peg-in-hole assembly can be completed successfully.Note to Practitioners—This paper proposes a novel contact force sensing method during the inserting operation. Compared with the traditional contact force sensing method, this paper adopts the force sensor integrated into the micromanipulator instead of commercial force sensor to detect the contact force between two parts. To ensure the assembling precision, the theoretical analysis is conducted to calculated the admissible contact force to avoid the sliding and rotating of the gripped micro part during assembling. This work efficiently simplifies the contact force sensing and control process, where complex calibration process needn’t to be carried out to eliminate the influence of the mass of the micromanipulator on the testing results. In addition, the assembling costs are reduced by replacing commercial force sensors with strain gauges.

We study the task of learning entity and relation embeddings in knowledge graphs for predicting missing links. Previous translational models on link prediction make use of translational properties but lack enough expressiveness, while the convolution neural network based model (ConvE) takes advantage of the great nonlinearity fitting ability of neural networks but overlooks translational properties. In this paper, we propose a new knowledge graph embedding model called ParamE which can utilize the two advantages together. In ParamE, head entity embeddings, relation embeddings and tail entity embeddings are regarded as the input, parameters and output of a neural network respectively. Since parameters in networks are effective in converting input to output, taking neural network parameters as relation embeddings makes ParamE much more expressive and translational. In addition, the entity and relation embeddings in ParamE are from feature space and parameter space respectively, which is in line with the essence that entities and relations are supposed to be mapped into two different spaces. We evaluate the performances of ParamE on standard FB15k-237 and WN18RR datasets, and experiments show ParamE can significantly outperform existing state-of-the-art models, such as ConvE, SACN, RotatE and D4-STE/Gumbel.

In spite of various observation tools, tongue shapes are still scarce resource in reality. Autoencoder, a kind of deep neural networks (DNN), performs well on data reduction and pattern discovery. However, since autoencoder usually needs large scale data in training, challenges exist for traditional autoencoder to obtain tongues' motion patterns only from tens or hundreds of available tongue shapes. To overcome this problem, we propose a two-steps autoencoder, where we first construct a stacked denoising autoencoder (dAE) to learn the essential presentation of the tongue shapes from their possible deformations; then an additional autoencoder with small number of hidden units is added upon the previous stacked autoencoder, and used for dimensionality reduction. Experiments run on 240 vowels' tongue shapes obtained from Chinese speakers' pronunciation X-ray films, and the proposed model is compared with traditional dAE and the classical principal component analysis (PCA) on dimensionality reduction and reconstruction in details. Results validate the performance of the proposed tongue model.

Pitch is an important characteristic of speech and is useful for many applications. However, it is still challenging to estimate pitch in strong noise. In this paper, we propose a joint training approach to determinate pitch. First, a Bidirectional Long Short-Term Memory Recurrent Neural Networks (BLSTMRNN) is trained to map the noisy to clean speech features. Second, the pitch estimation is also a BLSTM-RNN model. The feature mapping neural network serves as a noise normalization module aiming at explicitly generating the clean features which are easier to estimate pitch by the following neural network. BLSTM-RNN is trained on sequential frame-level features and capable of learning temporal dynamics. We also propose to take into account bottleneck features for pitch estimation. The experimental results show that the proposed method can obtain accurate pitch estimation and they show good generalization ability to new speakers and noisy conditions. The proposed approach also significantly outperforms other state-of-the-art pitch estimation algorithms.

Real-time magnetic resonance imaging (rtMRI) is becoming a practical tool in speech production research and language pathology observation. It is still a challenge to extract the tongue contour accurately in rtMRI sequences, since tongue is a soft tissue and often touches other organs such as lips and upper mandible. This paper proposes a novel semiautomatic tongue contour extraction method from rtMRI sequences. The initial boundary image is obtained by combined multi-directional Sobel operators in tongue movement region; then a boundary intensity map is constructed to find the most probable tongue contour points by searching for the optimal boundary route with Viterbi algorithm; finally the tongue contour is obtained using B-Spline approximation. The proposed method could obtain accurate tongue contour from rtMRI sequences, even in the cases that some parts of tongue touch other organs. Experiments demonstrate the robustness of the proposed method.

A high machining precision is the most important goal for the fabrication of three dimensional nanometer scale structure. The force-based atomic force microscope (AFM) can reach a very high precision in the structure machining, but the machining depth is difficult to predict for different sample material or machining parameters. The nano machining system in this paper adopts traditional displacement-based machining method, aiming at a few hundred nanometers machining depth. The machining tool (a probe) is kept motionless during the structure machining, so the structure size is determined by a high precise x/y/z three degree of freedoms (DOF) stage. A probe suspension with bridge-type amplification mechanism is designed and optimized, through the theoretical analysis and finite element analysis (FEA), the amplification factor, Z-direction stiffness and first natural frequency of the probe suspension are about 5, 5000N/m, 130HZ, respectively.

In spite of the development of new image techniques, X-ray remains an important technique in studying speech production phenomena. In this study, we propose an automatic contour extraction method of tongue in X-ray videos. At first, we take a region gradient based edge-detector to find the initial boundary points. As X-ray is high noise image and tongue is frequently occluded by teeth, there are high ratio outliers in the initial boundary point set. To solve this problem, we propose a cluster based point-to-point distance ratio filter to remove the outliers, which greatly reduces the iteration times of later RANSAC and B-Spline approximation for the final boundary points. Our method is nearly full-automatic, and obtains tongue's accurate contour. The experiments show that the proposed method could be an effective tool for tongue's continuous motion analysis.

This paper presents experimental analysis of laser interferometry-based closed-loop robust motion tracking control for flexure-based four-bar micro/nano manipulator. To enhance the accuracy of micro/nano manipulation, laser interferometry realized robust motion tracking control is established with the experimental facility. This paper contains brief discussions about the error sources associated with the laser interferometry-based sensing and measurement technique, along with detailed error analysis and estimation. Comparative error analysis of capacitive position sensor-based system and laser interferometry-based system is also presented. Robust control demonstrates high precision and accurate motion tracking of the four-bar flexure-based mechanism. The experimental results demonstrate precise motion tracking, where resultant closed-loop position tracking error is of the order of ± 20 nm, and a steady-state error of about ±10 nm. With the experimental study and error analysis, we offer evidence that the laser interferometry-based closed-loop robust motion tracking control can minimize positioning and tracking errors during dynamic motion.

This paper presents the dynamic modelling and performance evaluation methodologies of a flexure-base mechanism for ultra-precision grinding operations. The mechanical design of the mechanism is briefly described. A piezoelectric actuator is used to drive the moving platform. A flexure-based structure is utilized to guide the moving platform and to provide preload for the piezoelectric actuator. By simplifying the Hertzian contact as a linear spring and damping component, a bilinear dynamic model is developed to investigate the dynamic characteristics of the flexure-based mechanism. Based on the established model, the separation phenomenon of the moving platform from the piezoelectric actuator is analyzed. The influence of the control voltage on the maximum overshoot is also investigated. The slope and cycloidal command signals are used to reduce and/or avoid the overshoot of such flexure-based mechanism under step command signal actuation condition. The effects of the rising time of the command signals on the maximum overshoot and the settling time are studied.

This paper presents the forward kinematics and stiffness estimation methodologies of a flexure-based five-bar micro-manipulator. The mechanical design of the micro-manipulator is firstly descried. Base on the configuration of the proposed flexure-based micro-manipulator, the whole system has been divided into a five-bar parallel mechanism and an amplification mechanism. Two mathematical expressions describing the path traced by the tips of two passive links connected to each other are obtained. The Cartesian coordinates of the end-effector attached to one of the passive links is obtained according to the geometric relationship. The amplification factor of the lever mechanism is also derived based on the analytical solution of the four-bar linkage. By simplifying the flexure hinge as an ideal revolution joint with a linear torsional spring, the stiffness of the compliant five-bar mechanism is derived in both actuation and Cartesian spaces. It is noted that the stiffness of the compliant five-bar mechanism is positional dependant, and reaches up to infinite value at the singular configuration.