Extreme Learning Machine (ELM) is a training or learning algorithm for a single hidden layer feed-forward neural network with better performance.
The main goal of ELM is that the learning parameters of hidden nodes, including input weights and biases, are randomly assigned and need not be tuned, while the simple generalized inverse operation analytically determines the output weights. In the ELM algorithm, the only parameter need to be defined is the number of hidden nodes and produces a unique optimal solution. ELM outperforms traditional algorithm because it learns without iteration. The ELM yields better in terms of efficiency, accuracy, and stability.
The main advantage of the ELM is learning speed and generalization performance. Unified algorithm and ELM variants are the ELM learning algorithm, and improvements in ELM are online ELM and pruned ELM. Extreme Learning Machines are applied in many real-time learning tasks for classification, clustering regression, sparse approximation, compression, clustering, prediction, and also utilizes batch learning, sequential learning, and incremental learning.
The main applications of ELM are pattern recognition, forecasting, disease diagnosis, image processing, video processing, and medical signal processing. Various application areas of ELM are medical, chemistry, IoT, food industry, geography, robotics, transportation, economy, and many more. Future trends of ELM are extreme machine learning in big data with deep learning, ELM with random mechanisms, and combining ELM with biological learning.