We propose a coherent lightwave neural network system whose behavior can be controlled by the lightwave carrier frequency. We utilize the dependence of homodyne interference on the lightwave frequency by choosing the optical path lengths of the signal and the reference to be slightly different. The synaptic weights consist both of the optical path difference and variable delay realized by a spatial light modulator. A frequency-dependent learning is performed by a novel complex-valued Hebbian rule associated with the carrier frequency. Experiments demonstrate that the associative memory recalls different complex-amplitude patterns according to the carrier frequency variation successfully.
Lightwave has attractive characteristics such as spatial parallelism, temporal rapidity in signal processing, and frequency band vastness. In particular, the vast carrier frequency bandwidth promises novel information processing. In this paper, we propose a novel optical logic gate that learns multiple functions at frequencies different from one another, and analyze the frequency-domain multiplexing ability in the learning based on complex-valued Hebbian rule. We evaluate the averaged error function values in the learning process and the error probabilities in the realized logic functions. We investigate optimal learning parameters as well as performance dependence on the number of learning iterations and the number of parallel paths per neuron. Results show a trade-off among the learning parameters such as learning time constant and learning gain. We also find that when we prepare 10 optical path differences and conduct 200 learning iterations, the error probability completely decreases to zero in a three-function multiplexing case. However, at the same time, the error probability is tolerant of the path number. That is, even if the path number is reduced by half, error probability is found almost zero. The results can be useful to determine neural parameters for future optical neural network systems and devices that utilize the vast frequency bandwidth for frequency-domain multiplexing.
A coherent optical neural network is proposed that has the learning ability to achieve desirable phase values in the frequency domain. It is composed of multiple optical-path differences whose lengths are different from one another. The system learns a phase value at each discrete position in the frequency domain by obeying the complex-valued Hebbian rule. The learning curve also agrees with theoretical evolution.
We propose an adaptive logic circuit whose function can be controlled by optical carrier frequency modulation. The circuit learns the desired functions by adjusting the delay time at a spatial light modulator with a complex-valued Hebbian learning rule. After the learning, the circuit can switch its function all at once. A high degree of mechanical stability is achieved by spatial phase-difference coding. Two orthogonal phase components are detected in parallel spatially. Experiments demonstrate that the system works as an AND circuit at a certain frequency and as an XOR at another. The proposal will enhance the design of optical plastic cell architectures.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.