Approximate Computation for Big Data Analytics

Over the past a few years, research and development has made significant progresses on big data analytics with the supports from both governments and industries all over the world, such as Spark, IBM Watson and Google AlphaGo. A fundamental issue for big data analytics is the efficiency, and various advances towards attacking these issues have been achieved recently, from theory to algorithms to systems. In this talk, we shall present the idea of approximate computation for efficient and effective big data analytics: query approximation and data approximation, based on our recent research experiences. Different from existing approximation techniques, the approximation computation that we are going to introduce does not necessarily ask for theoretically guaranteed approximation solutions, but asks for sufficiently efficient and effective solutions in practice.