高维混合数据聚类是数据挖掘领域的研究热点和难点之一，广泛应用于互联网知识发现和管理决策过程。近年来，深度学习在图像、文字、声音等非结构化数据挖掘任务上的成功应用，为针对结构化高维混合数据的聚类研究提供了新思路。本项目以完善高维混合数据聚类方法为目标，引入深度学习理论，研究高维混合数据融合特征表示及聚类相关问题。主要研究内容包括：.（1）面向分类属性数据的深度无监督特征学习方法。利用深度学习模型训练分类属性数据，学习其潜在低维特征，并为后续基于深度学习的研究奠定基础。.（2）基于深度学习的混合数据融合特征表示及缺失数据处理方法。考虑不同类型属性间内在联系，学习高维混合数据的低维融合特征表示，并进一步给出兼顾缺失数据的特征学习方法。.（3）基于深度学习的高维混合数据聚类算法及有效性评价方法。同时考虑高维混合数据处理、缺失数据处理和不需要先决知识或参数的要求，优化高维混合数据聚类挖掘效果。

High dimensional and mixed-type data clustering analysis is one of the hot and difficult topics in data mining nowadays, and has a wide range of applications in internet knowledge discovery and management decision support. In recent years, the successful application of deep learning in unstructured data mining tasks such as images, texts, and sounds has provided fresh thought for clustering research on structured high dimensional and mixed-type data. This project aims at perfecting high dimensional and mixed-type data clustering method, and introduces deep learning theory to research on the related problems of fused feature representation and clustering of high dimensional and mixed-type data. The main contents of the research include:.(1) Deep unsupervised feature learning method for categorical data. The deep learning model is used to train categorical data to study its latent and low dimensional features and lay the foundation for the further research based on deep learning..(2) Mixed-type data fused feature representation and missing data processing method based on deep learning. Learn the low dimensional and fused feature representation of high dimensional and mixed-type data considering the intrinsic relationship between different types of attributes, and further give the feature learning method that takes into account the missing data..(3) High dimensional and mixed-type data clustering algorithm and cluster validation method based on deep learning, considering the requirements of high dimensional and mixed-type data processing, missing data processing and no prior knowledge or parameters, to optimize the clustering effectiveness of high dimensional and mixed-type data.