Upload croissant_rai.json

croissant_rai.json

@@ -0,0 +1,343 @@
+{
+  "@context": {
+    "@language": "en",
+    "@vocab": "https://schema.org/",
+    "arrayShape": "cr:arrayShape",
+    "citeAs": "cr:citeAs",
+    "column": "cr:column",
+    "conformsTo": "dct:conformsTo",
+    "cr": "http://mlcommons.org/croissant/",
+    "data": {
+      "@id": "cr:data",
+      "@type": "@json"
+    },
+    "dataBiases": "cr:dataBiases",
+    "dataCollection": "cr:dataCollection",
+    "dataType": {
+      "@id": "cr:dataType",
+      "@type": "@vocab"
+    },
+    "dct": "http://purl.org/dc/terms/",
+    "extract": "cr:extract",
+    "field": "cr:field",
+    "fileProperty": "cr:fileProperty",
+    "fileObject": "cr:fileObject",
+    "fileSet": "cr:fileSet",
+    "format": "cr:format",
+    "includes": "cr:includes",
+    "isArray": "cr:isArray",
+    "isLiveDataset": "cr:isLiveDataset",
+    "jsonPath": "cr:jsonPath",
+    "key": "cr:key",
+    "md5": "cr:md5",
+    "parentField": "cr:parentField",
+    "path": "cr:path",
+    "personalSensitiveInformation": "cr:personalSensitiveInformation",
+    "recordSet": "cr:recordSet",
+    "references": "cr:references",
+    "regex": "cr:regex",
+    "repeated": "cr:repeated",
+    "replace": "cr:replace",
+    "sc": "https://schema.org/",
+    "separator": "cr:separator",
+    "source": "cr:source",
+    "subField": "cr:subField",
+    "transform": "cr:transform",
+    "rai": "http://mlcommons.org/croissant/RAI/"
+  },
+  "@type": "sc:Dataset",
+  "distribution": [
+    {
+      "@type": "cr:FileObject",
+      "@id": "repo",
+      "name": "repo",
+      "description": "The Hugging Face git repository.",
+      "contentUrl": "https://huggingface.co/datasets/hcarrion/ControllabeGenDDI/tree/refs%2Fconvert%2Fparquet",
+      "encodingFormat": "git+https",
+      "sha256": "https://github.com/mlcommons/croissant/issues/80"
+    },
+    {
+      "@type": "cr:FileSet",
+      "@id": "parquet-files-for-config-default",
+      "containedIn": {
+        "@id": "repo"
+      },
+      "encodingFormat": "application/x-parquet",
+      "includes": "default/*/*.parquet"
+    }
+  ],
+  "recordSet": [
+    {
+      "@type": "cr:RecordSet",
+      "dataType": "cr:Split",
+      "key": {
+        "@id": "default_splits/split_name"
+      },
+      "@id": "default_splits",
+      "name": "default_splits",
+      "description": "Splits for the default config.",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "@id": "default_splits/split_name",
+          "dataType": "sc:Text"
+        }
+      ],
+      "data": [
+        {
+          "default_splits/split_name": "train"
+        }
+      ]
+    },
+    {
+      "@type": "cr:RecordSet",
+      "@id": "default",
+      "description": "hcarrion/ControllabeGenDDI - 'default' subset",
+      "field": [
+        {
+          "@type": "cr:Field",
+          "@id": "default/split",
+          "dataType": "sc:Text",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "fileProperty": "fullpath"
+            },
+            "transform": {
+              "regex": "default/(?:partial-)?(train)/.+parquet$"
+            }
+          },
+          "references": {
+            "field": {
+              "@id": "default_splits/split_name"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/image",
+          "dataType": "sc:ImageObject",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "image"
+            },
+            "transform": {
+              "jsonPath": "bytes"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/text",
+          "dataType": "sc:Text",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "text"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/file_name",
+          "dataType": "sc:Text",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "file_name"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/synthetic_type",
+          "dataType": "sc:Integer",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "synthetic_type"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/prompt_healthy_image_id",
+          "dataType": "cr:Int32",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "prompt_healthy_image_id"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/prompt_healthy_image_skin_tone",
+          "dataType": "cr:Int32",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "prompt_healthy_image_skin_tone"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/prompt_healthy_image_has_ruler",
+          "dataType": "sc:Boolean",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "prompt_healthy_image_has_ruler"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/prompt_disease",
+          "dataType": "sc:Text",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "prompt_disease"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/prompt_disease_malignancy",
+          "dataType": "cr:Float32",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "prompt_disease_malignancy"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/prompt_text_skin_tone",
+          "dataType": "cr:Float32",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "prompt_text_skin_tone"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/parametric_synthetic",
+          "dataType": "cr:Float32",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "parametric_synthetic"
+            }
+          }
+        },
+        {
+          "@type": "cr:Field",
+          "@id": "default/training_samples",
+          "dataType": "cr:Float32",
+          "source": {
+            "fileSet": {
+              "@id": "parquet-files-for-config-default"
+            },
+            "extract": {
+              "column": "training_samples"
+            }
+          }
+        }
+      ]
+    }
+  ],
+  "conformsTo": "http://mlcommons.org/croissant/1.1",
+  "name": "ControllabeGenDDI",
+  "description": "\n\t\n\t\t\n\t\tDataset Card for cgDDI\n\t\n\nThis repository is the official dataset of cgDDI: Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification which is currently under review. As this paper is not openly distributed (or on Arxiv), the method detail here will be minimal.\n\n\t\n\t\t\n\t\tDataset Details\n\t\n\nWe do not re-host previous artifacts, thus remember to fetch the original DDI dataset at https://ddi-dataset.github.io/ and the sDDI masks from… See the full description on the dataset page: https://huggingface.co/datasets/hcarrion/ControllabeGenDDI.",
+  "alternateName": [
+    "hcarrion/ControllabeGenDDI",
+    "cgDDI"
+  ],
+  "creator": {
+    "@type": "Person",
+    "name": "Héctor Carrión",
+    "url": "https://huggingface.co/hcarrion"
+  },
+  "keywords": [
+    "image-to-image",
+    "text-to-image",
+    "image-classification",
+    "English",
+    "apache-2.0",
+    "100K - 1M",
+    "parquet",
+    "Image",
+    "Text",
+    "Datasets",
+    "Dask",
+    "Croissant",
+    "Polars",
+    "🇺🇸 Region: US",
+    "medical"
+  ],
+  "license": "https://choosealicense.com/licenses/apache-2.0/",
+  "url": "https://huggingface.co/datasets/hcarrion/ControllabeGenDDI",
+  "version": "1.1",
+  "citeAs": "Carrión, H., et al. (2025). cgDDI: Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification. [Paper under review]",
+  "datePublished": "2025-04-29",
+  "url": "https://huggingface.co/datasets/hcarrion/ControllabeGenDDI",
+  "rai:dataCollection": "Base training dataset (DDI) was collected via the official download channels. Please see: https://ddi-dataset.github.io/ for direct download and ethical approvals (IRB protocols 36050 and 61146). DDI has been de-identified from patient records. We do not re-host DDI. Synthetic output dataset (cgDDI) was generated via parametric and non-parametric methods as described on the paper. We do not train on any imagery with personally identifiable information (PII), including faces, tattoos, and unique identifiers.",
+  "rai:dataCollectionType": [
+    "Synthetic medical imaging. ",
+    "Selection criteria for healthy imaging, lesion-mapping and parametric synthets was defined by reserchers in the medical AI field (the authros) who are not clinicians."
+  ],
+  "rai:dataBiases": [
+    "Skin-tone distribution: we attempt to aliviate un-even skin-tone distributions by measuring Imbalance Ratio (IR) where IR=1 signifies perfect balance. Healthy synthetics show an IR of 1.18, Lesion-map synthetics show an IR of 1.25 and Semantic synthetcis show an IR of 1.00. A potential bias in the data if not filtered by the end user could be disease-imbalance. However, we assure a minimum number of observations per disease. Selection and discard criteria was determined by two medical AI researchers (latino man, middle-eastern woman)."
+  ],
+  "rai:dataLimitations": [
+    "While our experiments demonstrate cgDDI improves classification performance and fairness, the semantic generation method requires sufficient training data for high-quality generation. This means many uncommon conditions in DDI cannot be parametrically generated to a satisfactory degree, thus our classification method trains on semantic synthetic diseases with more than 10 observations. Additionally, our approach provides generative control via prompting. However, instruction following is strongly influenced by sampling hyperparameters. Our parameter selection, while informed by preliminary experiments, was not exhaustively optimized due to computational costs. We address this through lesion-mapped synthetics, adding controllable rare disease augmentation in a non-parametric manner. Finally, while we manually review healthy samples, incorporating dermatologist review across all our subsets would measure dataset quality and provide information for filtering."
+  ],
+  "rai:personalSensitiveInformation": [
+    "This work builds exclusively on the publicly available DDI dataset, which was collected under Stanford IRB protocol numbers 36050 and 61146 with all clinical images de-identified (no faces, tattoos, or other Personally Identifiable Information (PII)). No new human subjects were enrolled."
+  ],
+  "rai:dataAnnotationProtocol": "Labelling or annotation was not necessary for the creation fo the cgDDI dataset as DDI is has been previously annotated via biopsy and consensus via two board-certified dermatologists who had private acess to full patient records. Manual review and discard criteria was performed by the paper authros with full protocol details available on Appendix section C.",
+  "rai:annotatorDemographics": [
+    "Manual review and discard criteria was performed by two researchers in the Medical AI field (paper authros) with the following demographics: one hispanic latino man and one middle-eastern woman."
+  ],
+  "rai:dataUseCases": [
+    "Training machine learning models for dermatological malignancy or condition classification",
+    "Validation and testing of AI systems for skin cancer detection",
+    "Research on fair and unbiased medical AI systems",
+    "Data augmentation for underrepresented skin-tones and rare malignancies."
+  ],
+  "rai:dataSocialImpact": "This dataset aims to improve dermatological AI systems that work equitably across different skin tones and demographics, potentially improving access to dermatological screening in underserved communities.",
+  "rai:dataReleaseMaintenancePlan": [
+    "Dataset will be maintained by the authors with further reviews for quality. Version updates will be released as needed with clear documentation of changes.The authros welcome community feedback, especially from board-certified dermatologists."
+  ]
+}