Computing device that performs a method for diagnosing properties of data and a system comprising the computing device
According to an embodiments of the present disclosure, a method comprising: at an electronic device with one or more processors, obtaining a data set; identifying, based on the data set, a first data point set on a first embedding space, wherein each data point included in the first data point set corresponds to each data included in the data set; identifying a modified first data point set on the first embedding space based on the first data point set by adjusting a property associated with a distribution of the first data point set, wherein the modified first data point set includes at least one modified data point which is not included in the first data point set; and providing a Modified Image of Data (MIOD) by representing the modified first data point set on an imaging space may be provided.
1 . A computing device, comprising:
a memory; and
at least one processor configured to operating based on at least one instruction stored in the memory;
wherein at least one processor is configured to:
obtain a data set,
identify a first data point set on a first embedding space by obtaining at least one feature of the data set from at a layer of a first trained model, wherein a first property of the data set is associated with dimensionality of the first data point set,
identify a second data point set on a second embedding space by obtaining at least one feature of the data set from at a layer of a second trained model, wherein a first modified property of the data set is associated with dimensionality of the second data point set, and wherein the first modified property of the data set is different from the first property of the data set;
obtain a second property of the data set based on-the first data point set-by obtaining output corresponding to the second property from a layer of a third trained model, wherein the second property is associated with a density of at least one data point included in the first data point set;
obtain a second modified property of the data set based on the second data point set by obtaining output corresponding to the second modified property from a layer of the third trained model, wherein the second modified property is associated with a density of at least one data point included in the second data point set, and wherein the second property and the second modified property are different; and
generate a diagnostic information including a property information and an Image of Data, wherein the diagnostic information includes an information on a quality of data, wherein the information on the quality of data is obtained by computing a metric that indicates the quality of the data based on the second property of the data set, the second modified property of the data set and wherein the Image of Data visually represents the data point set.
2 . The computing device of claim 1 , wherein the property information includes a third property information representing a third property of the data set which corresponds to a bias value of at least portion of the data set.
3 . The computing device of claim 1 , wherein the diagnostic information includes a comparison information representing a difference between the Image of Data and a Modified Image of Data, and wherein the Modified Image of Data visually represents the second data point set.
4 . The computing device of claim 1 , wherein the diagnostic information includes an information on data modification.
5 . The computing device of claim 1 , wherein at least one processor is further configured to:
provide a diagnostic report including the diagnostic information and an additional information on the data set.
6 . The computing device of claim 1 , wherein each data point included in the first data point set corresponds to each data included in the data set.
7 . The computing device of claim 1 , wherein the data set comprises a first data of a first modality and a second data of a second modality.
8 . The computing device of claim 7 , wherein the first modality is associated with an image data, and the second modality is associated with an annotation labeled on the image data.
9 . A method comprising:
at a computing device with one or more processors electronically connected to a memory, obtaining a data set;
identifying a first data point set on a first embedding space by obtaining at least one feature of the data set from at a layer of a first trained model, wherein a first property of the data set is associated with dimensionality of the first data point set;
identifying a second data point set on a second embedding space by obtaining at least one feature of the data set from at a layer of a second trained model, wherein a first modified property of the data set is associated with dimensionality of the second data point set, and wherein the first modified property of the data set is different from the first property of the data set;
obtaining a second property of the data set based on the first data point set by obtaining output corresponding to the second property from a layer of a third trained model, wherein the second property is associated with a density of at least one data point included in the first data point set;
obtaining a second modified property of the data set based on the second data point set by obtaining output corresponding to the second modified property from a layer of the third trained model, wherein the second modified property is associated with a density of at least one data point included in the second data point set, and wherein the second property and the second modified property are different; and
generating a diagnostic information including a property information and an Image of Data, wherein the diagnostic information includes an information on a quality of data, wherein the information on the quality of data is obtained by computing a metric that indicates the quality of the data based on the second property of the data set, the second modified property of the data set and wherein the Image of Data visually represents the data point set.
10 . The method of claim 9 , wherein the property information includes a third property information representing a third property of the data set which corresponds to a bias value of at least portion of the data set.
11 . The method of claim 9 , further comprising:
providing a diagnostic report including the diagnostic information and an additional information on the data set.
12 . The method of claim 9 , wherein the diagnostic information includes an information on data modification.
13 . The method of claim 9 , wherein each data point included in the first data point set corresponds to each data included in the data set.
14 . The method of claim 9 , wherein the data set comprises a first data of a first modality and a second data of a second modality, and wherein the first modality is associated with an image data, and the second modality is associated with an annotation labeled on the image data.
15 . A non-transitory computer-readable storage medium, storing program instructions computer-executable on a computer to perform operations comprising:
obtaining a data set;
identifying a first data point set on a first embedding space by obtaining at least one feature of the data set from at a layer of a first trained model, wherein a first property of the data set is associated with dimensionality of the first data point set;
identifying a second data point set on a second embedding space by obtaining at least one feature of the data set from at a layer of a second trained model, wherein a first modified property of the data set is associated with dimensionality of the second data point set, and wherein the first modified property of the data set is different from the first property of the data set;
obtaining a second property of the data set based on the data point set by obtaining output corresponding to the second property from a layer of a third trained model, wherein the second property is associated with a density of at least one data point included in the first data point set;
obtaining a second modified property of the data set based on the second data point set by obtaining output corresponding to the second modified property from a layer of the third trained model, wherein the second modified property is associated with a density of at least one data point included in the second data point set, and wherein the second property and the second modified property are different; and
generating a diagnostic information including a property information and an Image of Data, wherein the diagnostic information includes an information on a quality of data, wherein the information on the quality of data is obtained by computing a metric that indicates the quality of the data based on the second property of the data set, the second modified property of the data set and wherein the Image of Data visually represents the data point set.