Motion velocity estimation from electroencephalography signals with extreme learning machine

Decoding motion velocity from electroencephalography (EEG) signals is important for brain-computer interface (BCI) research. However, no studies explore how to decode the velocity of complex motion. In this paper, we apply extreme learning machine (ELM) to explore how to decode the velocity of complex motion from EEG signals. We design a new experimental paradigm and analyze the effects of the number of hidden neuron nodes and frequency band on the decoding performance. This work lays a foundation of building accurate motion decoders from EEG signals to develop the BCI-based prostheses and rehabilitation systems.