In this paper, we propose a new model to predict the age and number of suspects through the feature modeling of historical data. We discrete the case information into values of 20 dimensions. After feature selection, we use 9 machine learning algorithms and Deep Neural Networks to extract the numerical features. In addition, we use Convolutional Neural Networks and Long Short- Term Memory to extract the text features of case description. These two types of features are fused and fed into fully connected layer and softmax layer. This work is an extension of our short conference proceeding paper. The experimental results show that the new model improved accuracy by 3% in predicting the number of suspects and improved accuracy by 12% in predicting the number of suspects. To the best of our knowledge, it is the first time to combine machine learning and deep learning in crime prediction.
In view of the weak generalization of traditional event recognition methods, the limitation of dependence on field knowledge of expert, the longer train time of deep neural network, and the problem of gradient dispersion, the neural network joint model, Conv-RDBiGRU, integrated residual structure was proposed. Firstly, text corpus is preprocessed by word segmentation and stop words processing and uses word embedding to form the matrix of word vectors. Then, local semantic features are extracted through convolution operation, and deep context semantic features are extracted through RDBiGRU. Finally, the learned features are activated by softmax function and the recognition results are output. The novelty of work is that we integrate residual structure into recurrent neural network and combine these methods and field of application. The simulation results show that this method improves precision and recall of Chinese emergency event recognition, and the F-value is better than other methods.
Aiming at the problem of poor portability of traditional event recognition methods, the need for a large number of learning features, and the poor interpretability of recurrent neural networks in different information features about degrees of importance, this paper proposes a Chinese abrupt event recognition method based CBiGRU-Att model. Firstly, the text corpus was preprocessed.Word2vec was used to generate word vectors and the local features of the word vectors were extracted by using the convolutional layer. Then the extracted features were used as the input of the BiGRU to obtain higher-order context features, and introduced the attention mechanism to weight feature. Finally, softmax function was used to activate the learned features and output the recognition results. Simulation results show that this method is superior to other methods in the precision and recall rate for Chinese abrupt event recognition.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.