Judgment prediction for legal cases has attracted much research efforts for its practice use, of which the ultimate goal is prison term prediction. While existing work merely predicts the total prison term, in reality a defendant is often charged with multiple crimes. In this paper, we argue that charge-based prison term prediction (CPTP) not only better fits realistic needs, but also makes the total prison term prediction more accurate and interpretable. We collect the first large-scale structured data for CPTP and evaluate several competitive baselines. Based on the observation that fine-grained feature selection is the key to achieving good performance, we propose the Deep Gating Network (DGN) for charge-specific feature selection and aggregation. Experiments show that DGN achieves the state-of-the-art performance.

Accurate dynamic parameters are required in the dynamic model–based robot motion controller designing. Experiment-based robot parameter identification is the only way to obtain them effectively, in which motion parameters and actuator torques are measured and used to estimate the actual dynamic model. To simplify the identification procedure and improve the identification precision, a step-by-step identification method was proposed and investigated in this article with dynamic parameters of wrist and elbow joint estimated separately considering the difference in inertia parameters between wrist joint and elbow joint of typical no-load 6-DOF robots. The effectiveness, simplification and high precision compared with the traditional one-step identification method were demonstrated through comparative experiments.

No abstract

In recent years, aerial manipulators consist of unmanned aerial vehicles and robotic manipulators have been widely utilized in aerial operations. The complex dynamic coupling effects between unmanned aerial vehicles and robotic manipulators will bring some issues to the motion control. Therefore, the article proposes a new control scheme for aerial manipulators. The proposed method includes three elements, that is, time-delay estimation, backstepping design, and nonsingular terminal sliding mode. The time-delay estimation technique is adopted to estimate the complex system dynamics and to bring a model-free feature of the system. With the backstepping design, the proposed control strategy can ensure the asymptotic stability of the closed-loop system by recursive procedure. To deal with the unmodelling dynamics and disturbances, and to assure finite-time convergence of the system states, the nonsingular terminal sliding mode is adopted. By combining three elements, the tracking performance of aerial manipulators is improved under unmodelling dynamics and disturbances. Global stability of closed-loop control system is analyzed using Lyapunov stability theory. Finally, comparative simulations are conducted, and the results show that the proposed controller has better performance than a conventional proportional–derivative controller or a nonsingular terminal sliding mode controller.

T)ACSA tasks, including aspect-category sentiment analysis (ACSA) and targeted aspectcategory sentiment analysis (TACSA), aims at identifying sentiment polarity on predefined categories. Incremental learning on new categories is necessary for (T)ACSA real applications. Though current multi-task learning models achieve good performance in (T)ACSA tasks, they suffer from catastrophic forgetting problems in (T)ACSA incremental learning tasks. In this paper, to make multi-task learning feasible for incremental learning, we proposed Category Name Embedding network (CNE-net). We set both encoder and decoder shared among all categories to weaken the catastrophic forgetting problem. Besides the origin input sentence, we applied another input feature, i.e., category name, for task discrimination. Our model achieved state-of-theart on two (T)ACSA benchmark datasets. Furthermore, we proposed a dataset for (T)ACSA incremental learning and achieved the best performance compared with other strong baselines.