In recent years, remarkable works on artificial intelligence from different perspectives have been proposed by researchers, and the works have made significant progress. However, it is still a challenging issue to design new theoretical algorithms and systems on artificial intelligence. In the past few years, the awardees have dedicated pioneering works by proposing the Broad Learning Network (BLS), which is effective and efficient (much less time-consuming) compared with multilayer perceptron and deep structures. In this pioneering work, they have proposed incremental learning algorithms which greatly reduce the training time for remodeling when new samples are entering or when training is not accurate enough. The awardees also proved the universal approximation theorem of BLS to ensure its theoretical basis in various practical applications. In addition, the awardees have proposed several fuzzy neural network models on the basis of BLS, which provide possible mathematical explanation for the deep model. Their models have been followed by many researchers in major universities and research institutions, the BLS model has also been applied to solve classification problems in various fields, such as image/video applications, image recognition, and time-series prediction problems.

Fig 1 Illustration of BLS and Its Incremental Learning Algorithms

Fig 2 Research and Applications of BLS

Fig 3 The Structure of Fuzzy BLS