Abstract

Neural networks (NN) and support vector machines (SVM) play key roles in machine learning and data analysis in the past 2-3 decades. However, it is known that these popular learning techniques face some challenging issues such as: intensive human intervene, slow learning speed, poor learning scalability. This talk will introduce a next generation of learning theory; the resultant biologically inspired learning technique referred to as Extreme Learning Machine (ELM) and its wide applications. ELM not only learns up to tens of thousands faster than NN and SVMs, but also provides unified implementation for regression, binary and multi-class applications. ELM not only produces good results for sparse datasets but also is efficient for large size of applications. From both theoretical and practical points of view, NN and SVM/LS-SVM may only produce suboptimal solutions to ELM. Our preliminary study also shows that ELM outperforms Deep Learning in both learning accuracy and learning speed (up to tens of thousands times faster). ELM is efficient in time series, online sequential, incremental applications. More and more researchers are studying ELM and its potential applications in face recognition, EEG signal processing, brain computer interface, medical image processing, bioinformatics, disease prediction/detection, object recognition, knowledge discovery, data privacy, security, image quality assessment, semantic web, hardware implementation, cloud computing, and many other industrial applications.