Abstract: Systems and methods for document classification are described. Embodiments of the present disclosure generate classification data for a plurality of samples using a neural network trained to identify a plurality of known classes; select a set of samples for annotation from the plurality of samples using an open-set metric based on the classification data, wherein the annotation includes an unknown class; and train the neural network to identify the unknown class based on the annotation of the set of samples.

Abstract: Systems and techniques for privacy preserving document analysis are described that derive insights pertaining to a digital document without communication of the content of the digital document. To do so, the privacy preserving document analysis techniques described herein capture visual or contextual features of the digital document and creates a stamp representation that represents these features without included the content of the digital document. The stamp representation is projected into a stamp embedding space based on a stamp encoding model generated through machine learning techniques capturing feature patterns and interaction in the stamp representations. The stamp encoding model exploits these feature interactions to define similarity of source documents based on location within the stamp embedding space. Accordingly, the techniques described herein can determine a similarity of documents without having access to the documents themselves.

Abstract: Systems and techniques for privacy preserving document analysis are described that derive insights pertaining to a digital document without communication of the content of the digital document. To do so, the privacy preserving document analysis techniques described herein capture visual or contextual features of the digital document and creates a stamp representation that represents these features without included the content of the digital document. The stamp representation is projected into a stamp embedding space based on a stamp encoding model generated through machine learning techniques capturing feature patterns and interaction in the stamp representations. The stamp encoding model exploits these feature interactions to define similarity of source documents based on location within the stamp embedding space. Accordingly, the techniques described herein can determine a similarity of documents without having access to the documents themselves.

Abstract: Systems and techniques for privacy preserving document analysis are described that derive insights pertaining to a digital document without communication of the content of the digital document. To do so, the privacy preserving document analysis techniques described herein capture visual or contextual features of the digital document and creates a stamp representation that represents these features without included the content of the digital document. The stamp representation is projected into a stamp embedding space based on a stamp encoding model generated through machine learning techniques capturing feature patterns and interaction in the stamp representations. The stamp encoding model exploits these feature interactions to define similarity of source documents based on location within the stamp embedding space. Accordingly, the techniques described herein can determine a similarity of documents without having access to the documents themselves.

Abstract: Systems and methods for restoring the appearances of scans of damaged physical documents. Ink bleed is removed and/or ink added to portions of a scanned image based on determining an ink bleed model by analyzing colors of pixels in the scanned image. Gaps in strokes are reconstructed based on analyzing pixel color at multiple angles around individual pixels in the scanned image to determine whether the individual pixels are part of a stroke. The appearance of the scanned image is also enhanced by comparing pixels that are not already close to a background color or ink color with other nearby pixels and, based on the nearby pixels, adjusting colors of the pixels that are not already close to the background color or ink color. These techniques are used individually or in combination to improve the appearance of the scanned image.

Abstract: Systems and methods for restoring the appearances of scans of damaged physical documents. Ink bleed is removed and/or ink added to portions of a scanned image based on determining an ink bleed model by analyzing colors of pixels in the scanned image. Gaps in strokes are reconstructed based on analyzing pixel color at multiple angles around individual pixels in the scanned image to determine whether the individual pixels are part of a stroke. The appearance of the scanned image is also enhanced by comparing pixels that are not already close to a background color or ink color with other nearby pixels and, based on the nearby pixels, adjusting colors of the pixels that are not already close to the background color or ink color. These techniques are used individually or in combination to improve the appearance of the scanned image.

Abstract: Apparatus, systems, and methods operate to obtain data from a first array constructed from a directed acyclic graph formed as a prefix tree having key strings associated with a plurality of interconnected nodes, including branch nodes coupled via branches to sibling nodes and child nodes. Reference numbers are assigned to nodes in a monotonic progression as the prefix tree is traversed along the plurality of nodes. Sibling nodes are assigned reference numbers before child nodes, and child nodes are assigned reference numbers according to the order of appearance of key string characters. The first array comprises the key strings ordered according to the reference numbers. A second array can be formed as a linear searchable index derived from data in the first array, with elements of the second array comprising the reference numbers. Additional apparatus, systems, and methods are disclosed.

Abstract: Apparatus, systems, and methods operate to obtain data from a first array constructed from a directed acyclic graph formed as a prefix tree having key strings associated with a plurality of interconnected nodes, including branch nodes coupled via branches to sibling nodes and child nodes. Reference numbers are assigned to nodes in a monotonic progression as the prefix tree is traversed along the plurality of nodes. Sibling nodes are assigned reference numbers before child nodes, and child nodes are assigned reference numbers according to the order of appearance of key string characters. The first array comprises the key strings ordered according to the reference numbers. A second array can be formed as a linear searchable index derived from data in the first array, with elements of the second array comprising the reference numbers. Additional apparatus, systems, and methods are disclosed.