Information processing device and information processing method
It is possible to improve estimation accuracy with regard to data in which significance is attached to a relative phase. Provided is an information processing device including an estimation unit configured to estimate a status by using a neural network. The neural network includes a first complex-valued neural network to which complex data is input, a phase difference computation layer from which phase difference for each element between a plurality of sets with regard to the complex data is output, and a second complex-valued neural network from which complex data is output on the basis of the phase difference.
1. An information processing device, comprising
a processor; and
a non-transitory storage medium containing program instructions stored thereon, execution of which by the computing device causes the information processing device to provide functions of:
an estimation unit configured to estimate a status by using a neural network,
wherein the neural network includes
a first complex-valued neural network to which complex data is input,
a phase difference computation layer to which an output of the first complex-valued neural network is input, the phase difference computation layer including a plurality of sets of complex-valued neurons within a same layer, and being configured to calculate a phase difference between any two of the plurality of sets, and
a second complex-valued neural network from which complex data is output on the basis of the phase difference.
2. The information processing device according to claim 1 , wherein
the complex-valued neurons in one of said any two of the plurality of sets respectively correspond to the complex-valued neurons in the other of said any two of the plurality of sets; and
the phase difference computation layer outputs the phase difference on the basis of a conjugate product of corresponding complex-valued neurons between the sets.
3. The information processing device according to claim 1 ,
wherein the phase difference computation layer divides complex-valued neurons belonging to the same layer into at least two or more groups to thereby obtain the plurality of sets.
4. The information processing device according to claim 1 , wherein the first complex-valued neural network is a complex-valued convolutional neural network.
5. The information processing device according to claim 4 , wherein the complex-valued convolutional neural network performs a convolution process by using a pair of convolution filters having structures different from each other.
6. The information processing device according to claim 5 , wherein the convolution filter has a real-valued weight.
7. The information processing device according to claim 1 , wherein the neural network further includes a transformation layer configured to transform complex-valued data output from the second complex-valued neural network into real-valued data.
8. An information processing method for an information processing device that includes
a processor, and
a non-transitory storage medium containing program instructions stored thereon, execution of which by the computing device causes the information processing device to provide functions of an estimation unit configured to estimate a status by using a neural network having a first complex-valued neural network, a phase difference computation layer and a second complex-valued neural network,
the method comprising:
inputting, by the processor, complex data to the first complex-valued neural network;
inputting, by the processor, an output of the first complex-valued neural network to the phase difference computation layer that includes a plurality of sets of complex-valued neurons within a same layer;
outputting, by the processor, from the phase difference computation layer, a phase difference between any two of the plurality of sets; and
outputting, by the processor, from the second complex-valued neural network, complex data on the basis of the phase difference.