Machine learning device, inference device, and non-transitory program recording medium
A machine learning method includes: a first principal component analysis step configured to perform principal component analysis on learning data so as to thereby generate two or more principal components of the learning data; a first image data generation step configured to generate virtual image data by assigning, from among the two or more principal components, the first principal component to the X coordinate of an XY plane and the second principal component to the Y coordinate of the XY plane; and a learning step configured to generate a trained model by performing machine learning using the image data as input data.
1 . A machine learning method comprising:
a first principal component analysis step configured to perform principal component analysis on learning data to generate two or more principal components of the learning data;
a first image data generation step configured to generate first virtual image data by assigning, from among the two or more principal components, the first principal component to an X coordinate of an XY plane, a second principal component to a Y coordinate of the XY plane, and a third principal component to a pixel value at each point on the XY plane; and
a learning step configured to generate a first trained model by performing machine learning using the generated first virtual image data as input data.
2 . The machine learning method according to claim 1 ,
wherein the first image data generation step is further configured to assign each of the fourth and subsequent principal components to pixel value at each point on the XY plane and a color component value of a color space at each point on the XY plane of the first virtual image data.
3 . The machine learning method according to claim 2 , comprising: a noise application step configured to apply noise to points around a point on the XY plane on which the virtual image data is generated.
4 . The machine learning method according to claim 1 , comprising: a noise application step configured to apply noise to points around a point on the XY plane on which the virtual image data is generated.
5 . An inference method comprising:
a second principal component analysis step configured to perform principal component analysis on observation data to generate two or more principal components of the observation data;
a second image data generation step configured to generate second virtual image data by assigning, from among the two or more principal components, a first principal component to an X coordinate of an XY plane, the second principal component to a Y coordinate of the XY plane, and a third principal component to a pixel value at each point on the XY plane; and
an inference step configured to perform inference using a trained model using the generated second virtual image data as input data.
6 . A non-transitory program recording medium comprising a trained model generated by:
performing principal component analysis on learning data to generate two or more principal components of the learning data;
generating first virtual image data by assigning, from among the two or more principal components, a first principal component to an X coordinate of an XY plane, a second principal component to a Y coordinate of the XY plane, and a third principal component to a pixel value at each point on the XY plane; and
generating a first trained model by performing machine learning using the generated first virtual image data as input data.