Learning to hash for big data: Current status and future trends

With the rapid development of information technology, explosion of data has occurred in most areas, which means that we have entered the era of big data. Big data has become one of the most important national strategic resources owing to its wide application in a large variety of areas. As a result, research in both academia and industry has focused greatly on big data processing, including storage, management, and analysis. Because the ultimate goal of big data processing is to mine value from big data, in which machine learning plays a key role, big data machine learning (BDML) has become one of the core directions for big data research. By representing the data as binary code, learning to hash (LH) can dramatically reduce the storage and communication cost, thereby improving the efficiency and scalability of BDML systems. Furthermore, LH can also alleviate the curse of dimensionality in BDML systems. Hence, LH has become a hot research topic in machine learning and BDML. This paper gives a brief introduction to LH.