Search

Neural machine translation (NMT) is a data-driven machine translation approach that has proven its superiority in large corpora, but it still has much room for improvement when the corpus resources are not abundant. This work aims to improve the translation quality of Traditional Mongolian-Chinese (MN-CH). First, the baseline model is constructed based on the Transformer model, and then two different syntax-assisted learning units are added to the encoder and decoder. Finally, the encoder's ability to learn Traditional Mongolian syntax is implicitly strengthened, and the knowledge of Chinese-dependent syntax is taken as prior knowledge to explicitly guide the decoder to learn Chinese syntax. The average BLEU values measured under two experimental conditions showed that the proposed improved model improved by 6.706 (45.141–38.435) and 5.409 (41.930–36.521) compared with the baseline model. The analysis of the experimental results also revealed that the proposed improved model was still deficient in learning Chinese syntax, and then the Primer-EZ method was introduced to ameliorate this problem, leading to faster convergence and better translation quality. The final improved model had an average BLEU value increase of 9.113 (45.634–36.521) compared with the baseline model at experimental conditions of N = 5 and epochs = 35. The experiments showed that both the proposed model architecture and prior knowledge could effectively lead to an increase in BLEU value, and the addition of syntactic-assisted learning units not only corrected the initial association but also alleviated the long-term dependence between words. [ABSTRACT FROM AUTHOR]

Animated Graphics Interchange Format (GIF) images have become an important part of network information interaction, and are one of the main characteristics of analyzing social media emotions. At present, most of the research on GIF affection recognition fails to make full use of spatial-temporal characteristics of GIF images, which limits the performance of model recognition to a certain extent. A GIF emotion recognition algorithm based on ResNet-ConvGRU is proposed in this paper. First, GIF data is preprocessed, converting its image sequences to static image format for saving. Then, the spatial features of images and the temporal features of static image sequences are extracted with ResNet and ConvGRU networks, respectively. At last, the animated GIFs data features are synthesized and the seven emotional intensities of GIF data are calculated. The GIFGIF dataset is used to verify the experiment. From the experimental results, the proposed animated GIFs emotion recognition model based on ResNet-ConvGRU, compared with the classical emotion recognition algorithms such as VGGNet-ConvGRU, ResNet3D, CNN-LSTM, and C3D, has a stronger feature extraction ability, and sentiment classification performance. This method provides a finer-grained analysis for the study of public opinion trends and a new idea for affection recognition of GIF data in social media. [ABSTRACT FROM AUTHOR]

HotSpot is an algorithm that can directly mine association rules from real data. Aiming at the problem that the support threshold in the algorithm cannot be set accurately according to the actual scale of the dataset and needs to be set artificially according to experience, this paper proposes a dynamic optimization algorithm with minimum support threshold setting: S_HotSpot algorithm. The algorithm combines simulated annealing algorithm with HotSpot algorithm and uses the global search ability of simulated annealing algorithm to dynamically optimize the minimum support in the solution space. Finally, the Inner Mongolia sandstorm dataset is used for experiment while the wine quality dataset is used for verification, and the association rules screening indicators are set for the mining results. The results show that S_HotSpot algorithm can not only dynamically optimize the selection of support but also improve the quality of association rules as it is mining reasonable number of rules. [ABSTRACT FROM AUTHOR]

With the development of natural language processing and neural machine translation, the neural machine translation method of end-to-end (E2E) neural network model has gradually become the focus of research because of its high translation accuracy and strong semantics of translation. However, there are still problems such as limited vocabulary and low translation loyalty, etc. In this paper, the discriminant method and the Conditional Random Field (CRF) model were used to segment and label the stem and affixes of Mongolian in the preprocessing stage of Mongolian-Chinese bilingual corpus. Aiming at the low translation loyalty problem, a decoding model combining Convolution Neural Network (CNN) and Gated Recurrent Unit (GRU) was constructed. The target language decoding was performed by using the GRU. A global attention model was used to obtain the bilingual word alignment information in the process of bilingual word alignment processing. Finally, the quality of the translation was evaluated by Bilingual Evaluation Understudy (BLEU) values and Perplexity (PPL) values. The improved model yields a BLEU value of 25.13 and a PPL value of − 3 8. 1. The experimental results show that the E2E Mongolian-Chinese neural machine translation model was improved in terms of translation quality and semantic confusion compared with traditional statistical methods and machine translation models based on Recurrent Neural Networks (RNN). [ABSTRACT FROM AUTHOR]

The neural machine translation model based on long short-term memory (LSTM) has become the mainstream in machine translation with its unique coding–decoding structure and semantic mining features. However, there are few studies on the Mongolian and Chinese neural machine translation combined with LSTM. This paper mainly studies the preprocessing of Mongolian and Chinese bilingual corpus and the construction of the LSTM model of Mongolian morpheme coding. In the corpus preprocessing stage, this paper presents a hybrid algorithm for the construction of word segmentation modules. The sequence that has not been annotated is treated semantically and labeled by a combination of gated recurrent unit and conditional random field. In order to learn more grammar and semantic knowledge from Mongolian corpus, in the model construction stage, this paper presents the LSTM neural network model based on morpheme coding to construct the encoder. This paper also constructs the LSTM neural network decoder to predict the Chinese decode. Experimental comparisons of sentences of different lengths according to the construction model show that the model has improved translation performance in dealing with long-term dependence problems. [ABSTRACT FROM AUTHOR]

To realize the intelligent vehicle scheduling of public transportation, starting from the statistics data of GPS positioning and video surveillance, the optimization model of bus departure frequency was established. The BP neural network algorithm was used to predict the arrival time of the bus. The passenger flow of bus stops was forecasted, and according to the number of passengers on and off the bus collected by video, the number of passengers on different routes and stations at different time periods was predicted, and the prediction method was exponential smoothing. The bus departure frequency was arranged rationally, and through the establishment of objective function, the waiting time was reduced, the bus utilization rate was improved and the profitability of the bus company was increased. In the case of a variety of constraints, the final objective function was obtained by weighting, and the improved genetic algorithm was applied to obtain the optimal solution. The results showed that the bus frequency target was the minimum average waiting time of passengers and the bus average per trip passenger volume the maximum. To sum up, it is required to meet the standard deviation of the maximum section of every shift reach the minimum, the target with the maximum is transformed into the solution to the minimum, and the three comprehensively calculate the optimal calculation by the weighted sum. [ABSTRACT FROM AUTHOR]

The quality of meteorological observation data directly affects the weather forecast and the accuracy of climate prediction. The traditional quality control algorithm is not sensitive to the abnormal changes of the elements and can’t meet the needs of the quality control work. Therefore, based on the data mining algorithm, this paper further studied the quality control of meteorological data from two aspects of time correlation and factor correlation. Two different methods of quality control for meteorological observation data were proposed. One is the quality control method of time correlated meteorological observations based on the characteristics of chaos (potential trend and regularity) and the support vector machine algorithm. The other is the quality control method of factor correlated meteorological observations based on BP neural network and the characteristics of different elements. Combining the complementarity and relevance between the two methods, a set of comprehensive quality control scheme is set up. The experimental results show that the proposed scheme can effectively simulate the weather observation data and detect the anomaly value. [ABSTRACT FROM AUTHOR]