Figure | Learning with an all-optical neuron. (IMAGE)
Caption
Figure | Learning with an all-optical neuron. (a) The nonlinear PPLN chip and the programmable interferometer are linked together to form a functional optical neuron. Information is encoded onto light, transformed by the passive nonlinearity, and processed by interference to produce predictions. (b)–(d) The system successfully learns complex decision boundaries in standard classification tasks, achieving high accuracy. (e)–(g) When applied to multi-class data, the optical network forms clear separation regions and provides stable, interpretable predictions.
Credit
Wujie Fu et al.
Usage Restrictions
Credit must be given to the creator.
License
CC BY