The mixed reality conference system proposed in this paper is a robust, real-time video conference application software that makes up for the simple interaction and lack of immersion and realism of traditional video conference, which realizes the entire process of holographic video conference from client to cloud to the client. This paper mainly focuses on designing and implementing a video conference system based on AI segmentation technology and mixed reality. Several mixed reality conference system components are discussed, including data collection, data transmission, processing, and mixed reality presentation. The data layer is mainly used for data collection, integration, and video and audio codecs. The network layer uses Web-RTC to realize peer-to-peer data communication. The data processing layer is the core part of the system, mainly for human video matting and human-computer interaction, which is the key to realizing mixed reality conferences and improving the interactive experience. The presentation layer explicitly includes the login interface of the mixed reality conference system, the presentation of real-time matting of human subjects, and the presentation objects. With the mixed reality conference system, conference participants in different places can see each other in real-time in their mixed reality scene and share presentation content and 3D models based on mixed reality technology to have a more interactive and immersive experience.

Regular expression is important for many natural language processing tasks especially when used to deal with unstructured and semi-structured data. This work focuses on automatically generating regular expressions and proposes a novel genetic algorithm to deal with this problem. Different from the methods which generate regular expressions from character level, we first utilize byte pair encoder (BPE) to extract some frequent items, which are then used to construct regular expressions. The fitness function of our genetic algorithm contains multi objectives and is solved based on evolutionary procedure including crossover and mutation operation. In the fitness function, we take the length of generated regular expression, the maximum matching characters and samples for positive training samples, and the minimum matching characters and samples for negative training samples into consideration. In addition, to accelerate the training process, we do exponential decay on the population size of the genetic algorithm. Our method together with a strong baseline is tested on 13 kinds of challenging datasets. The results demonstrate the effectiveness of our method, which outperforms the baseline on 10 kinds of data and achieves nearly 9.7 percent improvement on F1 score on average. By doing exponential decay, the training speed is approximately 100 times faster than the methods without using exponential decay. In summary, our method possesses both effectiveness and efficiency, and can be implemented for the industry application.