Research on machine assisted text analysis follows the rapid development of digital media, and sentiment analysis is among the prevalent applications. Traditional sentiment analysis methods require complex feature engineering, and embedding representations have dominated leaderboards for a long time. However, the context-independent nature limits their representative power in rich context, hurting performance in Natural Language Processing (NLP) tasks. Bidirectional Encoder Representations from Transformers (BERT), among other pre-trained language models, beats existing best results in eleven NLP tasks (including sentence-level sentiment classification) by a large margin, which makes it the new baseline of text representation. As a more challenging task, fewer applications of BERT have been observed for sentiment classification at the aspect level. We implement three target-dependent variations of the BERT base model, with positioned output at the target terms and an optional sentence with the target built in. Experiments on three data collections show that our TD-BERT model achieves new state-of-the-art performance, in comparison to traditional feature engineering methods, embedding-based models and earlier applications of BERT. With the successful application of BERT in many NLP tasks, our experiments try to verify if its context-aware representation can achieve similar performance improvement in aspect-based sentiment analysis. Surprisingly, coupling it with complex neural networks that used to work well with embedding representations does not show much value, sometimes with performance below the vanilla BERT-FC implementation. On the other hand, incorporation of target information shows stable accuracy improvement, and the most effective way of utilizing that information is displayed through the experiment.INDEX TERMS Deep learning, neural networks, sentiment analysis, BERT.

No abstract

No abstract

This paper focused on three application problems of the traditional Deep Deterministic Policy Gradient(DDPG) algorithm. That is, the agent exploration is insufficient, the neural network performance is unsatisfied, the agent output fluctuates greatly. In terms of agent exploration strategy, network training algorithm and overall algorithm implementation, an improved DDPG method based on double-layer BP neural network is proposed. This method introduces fuzzy algorithm and BFGS algorithm based on Armijo-Goldstein criterion, improves the exploration efficiency, learning efficiency and convergence of BP neural network, increases the number of layers of BP neural network to improve the fitting ability of the network, and adopts periodic update to ensure the stable operation of the algorithm. The experimental results show that the deep learning network based on the improved DDPG algorithm has greatly improved the performance compared with the traditional method after multiple rounds of self-learning under variable working conditions. This study lays a theoretical and experimental foundation for the extended application of deep learning algorithm.

Hydraulic drive mode enables legged robots to have excellent characteristics, such as greater power-to-weight ratios, higher load capacities, and faster response speeds than other robots. Nowadays, highly integrated valve-controlled cylinder, called the hydraulic drive unit (HDU), is employed to drive the joints of these robots. However, there are some common problems in the HDU resulted from hydraulic systems, such as strong nonlinearity, asymmetry dynamic characteristics between positive and negative moving directions of the piston rod, and time-varying parameters. It is difficult to obtain the desired control performance by just using classical control methods such as the traditional PID control. In this paper, a position controller that combines fuzzy terminal sliding mode control (FTSMC) and time delay estimation (TDE) is proposed, in which FTSMC adopts a compound reaching law which combines the tangent function and the exponential reaching law. Moreover, the fuzzy control is introduced to adjust the parameters of the reaching law in real time to improve the adaptability of FTSMC. Based on FTSMC, the external uncertain disturbance of the HDU position control system is estimated by TDE, which ensures the simplicity of system modeling and the normal application of FTSMC. Finally, the control effects of the controller combining FTSMC and TDE are verified on the HDU performance test platform and the load simulation experiment platform. The experimental results show that the proposed controller greatly improves the system position control performance and has a strong disturbance rejection ability and a good adaptability under different working conditions. The above research results can provide an important reference and experimental basis for the inner loop of compliance control of legged robots.