Liquid state machine
A model of artificial Neural network. It consists of liquid filters and readouts. A liquid filter is a neural circuit that receives the input and produce ‘liquid states’, which is subsequently converted to an output by readout filters. It is naturally suitable to process real-time inputs.
References
- Liquid state machines: motivation, theory, and applications by Wolfgang Maass
- W. Maass, T. Natschlaeger, and H. Markram, Real-time computing without stable states: a new framework for neural computation based on perturbations, Neural Computation, 14(11), 2531-2560 (2002). pdf - the first paper
- Structured liquids in liquid state machines - A master thesis on LSM.