LEARNING APPARATUS, SIGNAL ESTIMATION APPARATUS, LEARNING METHOD, SIGNAL ESTIMATION METHOD, AND PROGRAM TO DEQUANTIZE
A neural network is learned that uses learning data including a low-bit signal obtained by quantizing a signal to a first number of quantization bits and a high-bit signal obtained by quantizing the signal to a second number of quantization bits larger than the first number of quantization bits, to receive as an input a low-bit input signal obtained by quantizing an input signal to the first number of quantization bits and output an estimated signal of a high-bit output signal obtained by quantizing the input signal to the second number of quantization bits. This neural network has a multilayer structure including an input layer and an output layer, and obtains and outputs an estimated signal of a high-bit output signal obtained by adding to a low-bit input signal a signal output from the output layer in response to the low-bit input signal being input to the input layer.
1 . A computer implemented method for learning a neural network, comprising:
learning a neural network using learning data including:
a low-bit signal obtained by quantizing a signal to a first number of quantization bits and
a high-bit signal obtained by quantizing the signal to a second number of quantization bits larger than the first number of quantization bits,
wherein the learnt neural network receives an input signal the low-bit input signal and outputs the high-bit output signal obtained as an estimated signal by quantizing the input signal to the second number of quantization bits, and wherein
the neural network includes a multilayer structure including an input layer and an output layer, and outputs the estimated signal by adding an output signal from the output layer to the low-bit input signal as a signal output in response to receiving the low-bit input signal as input to the input layer.
2 . The computer implemented method of claim 1 , wherein
the neural network comprises a multilayer structure including a layer that determines the output signal based on the input signal, and
the output signal is determined based at least on a product of a first column of a plurality of values obtained by performing a first convolution linear transformation processing on the input signal and a second column of a plurality of values obtained by performing a second convolution linear transformation processing on the input signal based on a value.
3 . The computer implemented method of claim 1 , wherein
the neural network includes a multilayer structure including a layer that determines the output signal based on the input signal, and
the output signal includes a product A×V′, where A represents a column corresponding to a product of a column K of a plurality of values obtained by performing convolution linear transformation processing W K on the input X and a column Q of a plurality of values obtained by performing convolution linear transformation processing W Q on the input signal, and V′ represents a column of a plurality of values obtained by performing convolution linear transformation processing W V on the input signal.
4 . The computer implemented method according to claim 1 , further comprising:
inputting another low-bit input signal quantized to the first number of quantization bits to the learnt neural network; and
obtaining and outputting another estimated signal quantized to the second number of quantization bits larger than the first number of quantization bits.
5 . A learning device comprising a processor configured to execute a method comprising:
learning a neural network that uses learning data including:
a low-bit signal obtained by quantizing a signal to a first number of quantization bits and
a high-bit signal obtained by quantizing the signal to a second number of quantization bits larger than the first number of quantization bits,
wherein the learnt neural network receives as an input a low-bit input signal obtained by quantizing an input signal to the first number of quantization bits and to output an estimated signal of a high-bit output signal obtained by quantizing the input signal to the second number of quantization bits, and wherein
the neural network comprises a multilayer structure including an input layer and an output layer, and the learnt network obtains and outputs an estimated signal of the high-bit output signal obtained by adding to the low-bit input signal a signal output from the output layer in response to the low-bit input signal being input to the input layer.
6 . A signal estimation device comprising a processor configured to execute a method comprising:
receiving as input a low-bit input signal obtained by quantizing an input signal to a first number of quantization bits to a neural network;
determining an estimated signal of a high-bit output signal obtained by quantizing the input signal to a second number of quantization bits larger than the first number of quantization bits; and
outputting the estimated signal.
7 . (canceled)
8 . The computer implemented method according to claim 1 , wherein the low-bit input signal corresponds to a music signal that is quantized based on an analog-digital conversion.
9 . The computer implemented method according to claim 1 , wherein the low-bit input signal corresponds to a sensor signal associated with a robot.
10 . The computer implemented method according to claim 1 , wherein the first number quantization is substantially close to 16 bits, and wherein the second number of quantization is substantially close to 24 bits.
11 . The computer implemented method according to claim 1 , wherein the multilayer structure includes at least three layers.
12 . The learning device according to claim 5 , wherein
the neural network comprises a multilayer structure including a layer that determines the output signal based on the input signal, and
the output signal is determined based at least on a product of a first column of a plurality of values obtained by performing a first convolution linear transformation processing on the input signal and a second column of a plurality of values obtained by performing a second convolution linear transformation processing on the input signal based on a value.
13 . The learning device according to claim 5 , wherein
the neural network includes a multilayer structure including a layer that determines the output signal based on the input signal, and
the output signal includes a product A×V′, where A represents a column corresponding to a product of a column K of a plurality of values obtained by performing convolution linear transformation processing W K on the input X and a column Q of a plurality of values obtained by performing convolution linear transformation processing W Q on the input signal, and V′ represents a column of a plurality of values obtained by performing convolution linear transformation processing W V on the input signal.
14 . The learning device according to claim 5 , the processor further configured to execute a method comprising:
inputting another low-bit input signal quantized to the first number of quantization bits to the learnt neural network; and
obtaining and outputting another estimated signal quantized to the second number of quantization bits larger than the first number of quantization bits.
15 . The learning device according to claim 5 , wherein the low-bit input signal corresponds to a music signal that is quantized based on an analog-digital conversion.
16 . The learning device according to claim 5 , wherein the low-bit input signal corresponds to a sensor signal associated with a robot.
17 . The learning device according to claim 5 , wherein the first number quantization is substantially close to 16 bits, and wherein the second number of quantization is substantially close to 24 bits.
18 . The learning device according to claim 5 , wherein the multilayer structure includes at least three layers.
19 . The signal estimation device according to claim 6 , wherein the neural network includes a multilayer structure including a layer that determines the output signal based on the input signal, and
the output signal includes a product A×V′, where A represents a column corresponding to a product of a column K of a plurality of values obtained by performing convolution linear transformation processing W K on the input X and a column Q of a plurality of values obtained by performing convolution linear transformation processing W Q on the input signal, and V′ represents a column of a plurality of values obtained by performing convolution linear transformation processing W V on the input signal.
20 . The signal estimation device according to claim 6 , wherein the low-bit input signal corresponds to a music signal that is quantized based on an analog-digital conversion.
21 . The signal estimation device according to claim 6 , wherein the low-bit input signal corresponds to a sensor signal associated with a robot.