Patents by Inventor Stefan Ravizza
Stefan Ravizza has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 11954612Abstract: A method includes receiving a first query by a computing device and assigning the first query to a plurality of cognitive engines, wherein each of the plurality of cognitive engines include different characteristics for processing data. The method also includes, responsive to receiving a response from each of the plurality of cognitive engines for the first query, comparing the received responses from the plurality of cognitive engines. The method also included responsive to determining a difference between a first response from a first cognitive engine and a second response from a second cognitive engine is above a predetermined threshold value, performing a response mediation process until the difference is below the predetermined threshold value. The method also includes selecting a first final response from the received responses for the first query and the second query and displaying the first final response to a user.Type: GrantFiled: September 5, 2017Date of Patent: April 9, 2024Assignee: International Business Machines CorporationInventors: Andrea Giovannini, Florian Graf, Stefan Ravizza, Tim U. Scheideler
-
Patent number: 11941000Abstract: An embodiment includes processing a dataset to generate a set of feature vectors that include a first feature vector corresponding to a first concept within a user's areas of interest and a second feature vector corresponding to a second concept within the user's areas of study. The embodiment identifies clusters of the feature vectors and identifies key features that most contribute to influencing the clustering algorithm. The embodiment selects the first feature vector in response to a user query, and then selects the second feature vector based on an overlap between key features of the first and second feature vectors and a degree of dissimilarity between the first and second concepts. The embodiment outputs a query response that includes the second concept. The embodiment also determines an effectiveness value based on sensor data indicative of a user action responsive to the outputting of the response to the query.Type: GrantFiled: April 16, 2021Date of Patent: March 26, 2024Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Shikhar Kwatra, Robert E. Loredo, Frederik Frank Flöther, Stefan Ravizza
-
Patent number: 11651276Abstract: A computer-implemented method for generating a group of representative model cases for a trained machine learning model may be provided. The method comprising determining an input space, determining an initial plurality of model cases, and expanding the initial plurality of model cases by stepwise modifying field values of the records representing the initial plurality of model cases resulting in an exploration set of model cases. Additionally, the method comprises obtaining a model score value for each record of the exploration set of model cases, continuing the expansion of the exploration set of model cases thereby generating a refined model case set, and selecting the records in the refined model case set based on relative record distance values and related model score values between pairs of records, thereby generating the group of representative model cases.Type: GrantFiled: October 31, 2019Date of Patent: May 16, 2023Assignee: International Business Machines CorporationInventors: Stefan Ravizza, Andrea Giovannini, Patrick Lustenberger, Frederik Frank Flöther, Thomas Pfeiffer
-
Patent number: 11641330Abstract: A method for personalizing a message between a sender and a receiver is provided. The method comprises semantically analyzing a communication history to form a knowledge graph, deriving formality level values using a first trained ML model, analyzing parameter values of replies to determine receiver impact score, and training a second ML system to generate a model to predict the receiver impact score value. The method also comprises selecting a linguistic expression in a message being drafted, determining an expression intent, modifying the linguistic expression based on the formality level and the expression intent to generate a modified linguistic expression, and testing whether the modified linguistic expression has an increased likelihood of a higher receiver impact score. The method also comprises repeating selecting the linguistic expression, determining the expression intent, modifying the linguistic expression, and testing until a stop criterion is met.Type: GrantFiled: August 6, 2020Date of Patent: May 2, 2023Assignee: International Business Machines CorporationInventors: Frederik Frank Flöther, Shikhar Kwatra, Patrick Lustenberger, Stefan Ravizza
-
Patent number: 11556825Abstract: Aspects of the present invention disclose a method for verifying labels of records of a dataset. The records comprise sample data and a related label out of a plurality of labels. The method includes one or more processors dividing the dataset into a training dataset comprising records relating to a selected label and an inference dataset comprising records with sample data relating to the selected label and all other labels out of the plurality of labels. The method further includes dividing the training dataset into a plurality of learner training datasets that comprise at least one sample relating to the selected label. The method further includes training a plurality of label-specific few-shot learners with one of the learner training datasets. The method further includes performing inference by the plurality of trained label-specific few-shot learners on the inference dataset to generate a plurality of sets of predicted label output values.Type: GrantFiled: November 26, 2019Date of Patent: January 17, 2023Assignee: International Business Machines CorporationInventors: Andrea Giovannini, Georgios Chaloulos, Frederik Frank Flother, Patrick Lustenberger, David Mesterhazy, Stefan Ravizza, Eric Slottke
-
Patent number: 11550827Abstract: A method and a related system for allocating target locations to optimize trajectories between several objects and the target locations may be provided. The method comprises capturing location data of the target locations as well as location and movement data of the objects, building a graph using the target locations as well as the location and movement data and integrating constraints into the graph. Furthermore, the method comprises determining for each of the several objects a desired target location using a first optimization system, thereby determining endpoints of a trajectory between each of the objects and its respective desired target location and selecting for each of the several objects an optimal path as the trajectory between the object and the desired target location, using a second optimization system, and taking into account movements of other objects along their trajectories.Type: GrantFiled: September 22, 2020Date of Patent: January 10, 2023Assignee: International Business Machines CorporationInventors: Stefan Ravizza, Shikhar Kwatra, Frederik Frank Flöther, Saurabh Yadav
-
Publication number: 20220335041Abstract: An embodiment includes processing a dataset to generate a set of feature vectors that include a first feature vector corresponding to a first concept within a user's areas of interest and a second feature vector corresponding to a second concept within the user's areas of study. The embodiment identifies clusters of the feature vectors and identifies key features that most contribute to influencing the clustering algorithm. The embodiment selects the first feature vector in response to a user query, and then selects the second feature vector based on an overlap between key features of the first and second feature vectors and a degree of dissimilarity between the first and second concepts. The embodiment outputs a query response that includes the second concept. The embodiment also determines an effectiveness value based on sensor data indicative of a user action responsive to the outputting of the response to the query.Type: ApplicationFiled: April 16, 2021Publication date: October 20, 2022Applicant: International Business Machines CorporationInventors: Shikhar Kwatra, Robert E. Loredo, Frederik Frank Flöther, Stefan Ravizza
-
Patent number: 11403328Abstract: A method for linking a first knowledge graph (KG) and a second KG in the presence of a third KG is provided. Content of nodes of the first KG is compared to nodes of the second KG. If a first KG node has a content relationship to a related second KG node, an edge identified by a tuple identifying the first KG and the first KG node and a tuple identifying the second KG and the second KG node is stored in a meta-layer KG. The method comprises comparing content of the nodes from the third KG with the content of nodes from the first and second KG, and in case relationships are identified, more complex tuples establishing this relationship in the meta-layer are stored. Finally, the method also comprises storing at least all nodes and edges of the meta-layer knowledge graph.Type: GrantFiled: March 8, 2019Date of Patent: August 2, 2022Assignee: International Business Machines CorporationInventors: Stefan Ravizza, Frederik Frank Flöther, Florian Graf, Erik Rueger, Andrea Giovannini
-
Patent number: 11379733Abstract: A method for event predictions is provided. The method includes receiving input data. The method further includes identifying an object in the input data with the identified object associated with a first node in a knowledge graph. The method further includes determining a second node of a first object event with the second node related to the first node in the knowledge graph. The method further includes contextualizing the identified input object with the first object event.Type: GrantFiled: July 10, 2019Date of Patent: July 5, 2022Assignee: International Business Machines CorporationInventors: Andrea Giovannini, Frederik Frank Flöther, Florian Graf, Stefan Ravizza, Erik Rueger
-
Patent number: 11379598Abstract: A method and a related system for controlling user access to a target node in a knowledge graph may be provided. The method comprises defining a knowledge graph structure limitation for a user, defining a node type depending on the number of edges connecting to the node, determining a condition for an access to the target node, based on the knowledge graph structure limitation relative to the start node and the node type of the target node, upon the user attempting, coming from a start node, to access the target node in the knowledge graph, and granting access to the target node based on the determination.Type: GrantFiled: February 28, 2018Date of Patent: July 5, 2022Assignee: International Business Machines CorporationInventors: Stefan Ravizza, Erik Rueger, Tim U. Scheideler, Peter Minig
-
Publication number: 20220164680Abstract: In an approach, a processor creates a multi-layered knowledge graph (KG), wherein a first layer is a core KG, a second layer has application-specific structured facts, and a third layer has individualized facts. A processor adapts weights in each layer of the multi-layered KG based on the individualized facts. A processor uses, as input data to the multi-layered KG, individual environmental data. A processor maps the input data to the multi-layered KG in a sequence of the first layer, the second layer, and the third layer. A processor selects, as relevant nodes in the first layer and the second layer, the relevant nodes lying on a selected path from the input data via the first layer, the second layer, and the third layer having the highest average weight value along the selected path. A processor outputs facts of the relevant nodes from the first layer and the second layer.Type: ApplicationFiled: November 24, 2020Publication date: May 26, 2022Inventors: Stefan Ravizza, Matthias Biniok, Frederik Frank Flöther, Patrick Lustenberger, David Ocheltree, Saurabh Yadav
-
Patent number: 11294958Abstract: A knowledge graph is divided into a plurality of sub-graphs, each sub-graph comprising a plurality of vertices and a plurality of edges. The knowledge graph is represented as a summary graph comprising for each of the sub-graphs a summary-graph vertex. A local sub-graph is generated as a copy of one of the sub-graphs together with a copy of a surrounding graph to the one of the sub-graphs. The content of the local sub-graph is modified. The local sub-graph is reintegrated, upon a reintegration trigger event, back into the knowledge graph, wherein a structure of the surrounding graph is used as a reintegration aid, by overlaying the structure and the knowledge graph, thereby identifying identical vertices of the surrounding structure and the knowledge graph as anchor points from where changes in the local sub-graph are reintegrated into the knowledge graph.Type: GrantFiled: February 28, 2020Date of Patent: April 5, 2022Assignee: Kyndryl, Inc.Inventors: Tim U. Scheideler, Erik Rueger, Frederik F. Flöther, Stefan Ravizza
-
Publication number: 20220092091Abstract: A method and a related system for allocating target locations to optimize trajectories between several objects and the target locations may be provided. The method comprises capturing location data of the target locations as well as location and movement data of the objects, building a graph using the target locations as well as the location and movement data and integrating constraints into the graph. Furthermore, the method comprises determining for each of the several objects a desired target location using a first optimization system, thereby determining endpoints of a trajectory between each of the objects and its respective desired target location and selecting for each of the several objects an optimal path as the trajectory between the object and the desired target location, using a second optimization system, and taking into account movements of other objects along their trajectories.Type: ApplicationFiled: September 22, 2020Publication date: March 24, 2022Inventors: Stefan Ravizza, Shikhar Kwatra, Frederik Frank Flöther, Saurabh Yadav
-
Publication number: 20220045975Abstract: A method for personalizing a message between a sender and a receiver is provided. The method comprises semantically analyzing a communication history to form a knowledge graph, deriving formality level values using a first trained ML model, analyzing parameter values of replies to determine receiver impact score, and training a second ML system to generate a model to predict the receiver impact score value. The method also comprises selecting a linguistic expression in a message being drafted, determining an expression intent, modifying the linguistic expression based on the formality level and the expression intent to generate a modified linguistic expression, and testing whether the modified linguistic expression has an increased likelihood of a higher receiver impact score. The method also comprises repeating selecting the linguistic expression, determining the expression intent, modifying the linguistic expression, and testing until a stop criterion is met.Type: ApplicationFiled: August 6, 2020Publication date: February 10, 2022Inventors: Frederik Frank Flöther, Shikhar Kwatra, Patrick Lustenberger, Stefan Ravizza
-
Patent number: 11183270Abstract: A system and machine-implemented method for sorting Next-Generation Sequencing (NGS) reads in O(n) time and space complexity that makes use low sparsity and nearly uniform distribution of the input array. The genome position field in the input array is used to determine the target position of the output array. Duplicate target positions due to n-fold coverage are handled by assigning either overflow buckets to each position or anterior assigning multiple target slots in the output array for each genome position depending on the distribution of reads over the genome and the resulting probability of hitting an already occupied slot. Once every tuple in the input array has been written to the output array, the output array in read through ascending order and each tuple is appended to the end of a final result array.Type: GrantFiled: December 7, 2017Date of Patent: November 23, 2021Assignee: International Business Machines CorporationInventors: Romeo Kienzler, Jenny Li, Stefan Mueck, Stefan Ravizza
-
Patent number: 11176429Abstract: A system for enhancing a classifier prediction in respect to underrepresented classes may be provided. A classifier system trained with training data to build a model is used for classifying unknown input data, and an evaluator engine adapted for a determination of an underrepresented class. Additionally, the system comprises an extractor engine adapted for an extraction of relating data from an additional source, and a similarity engine adapted for a selection of data sets out of the relating data wherein the similarity engine is also adapted for comparing features of the relating data and a representative data set for the underrepresented class. Finally, the system comprises a recursion unit adapted for triggering the evaluator engine, the extractor engine and the similarity engine treating selected data set as input data until the evaluator engine classifies the selected data set with a confidence level which is above a confidence threshold level.Type: GrantFiled: May 13, 2019Date of Patent: November 16, 2021Assignee: International Business Machines CorporationInventors: Markus Brandes, Frederik Frank Flöther, Andrea Giovannini, Florian Graf, Stefan Ravizza
-
Publication number: 20210350223Abstract: A method for performing an iteration of an output of a trained GAN may be provided. The method comprises receiving an object as input for the GAN, determining a set of features of the input by the generator adversarial network, generating, by the GAN, at least one modification to one feature of the set of features of the object, generating as output of the GAN the received object as a basis, and the generated modification building a modified object, capturing a feedback signal, and receiving the feedback signal as input by the GAN in a feedback loop for a next iteration. Moreover, the method comprises repeating the determination of a set of features, the generation of at least one modification, the generation of the output and the caption of the feedback signal in the next iteration, wherein as object the modified object is used as the object.Type: ApplicationFiled: May 7, 2020Publication date: November 11, 2021Inventors: Christian Eggenberger, Frederik Frank Flöther, Shikhar Kwatra, Stefan Ravizza
-
Patent number: 11120150Abstract: A computer-implemented method, system, and computer program product for dynamic access control to a node in a knowledge graph includes: structuring nodes of a knowledge graph into a plurality of hierarchically organized graph layers; assigning, to one or more users, an access right to a first node of the knowledge graph, the access right to the node selected from a plurality of access rights, where different types of users have different access rights; and assigning, to at least one user from the one or more users, an additional access right to a second node of the knowledge graph.Type: GrantFiled: February 26, 2020Date of Patent: September 14, 2021Assignee: International Business Machines CorporationInventors: Stefan Ravizza, Tim U. Scheideler, Florian Graf, Andrea Giovannini, Frederik Flöther, Erik Rueger
-
Patent number: 11086909Abstract: A method for partitioning a knowledge graph is provided. The method analyzes past searches and determines an access frequency of a plurality of edges. The method marks, as intermediate cluster cores, edges having the highest access frequencies, sorts the marked intermediate cluster cores according to their access frequencies, and selects a first cluster core having the highest access frequency. The method assigns first edges in a first radius around the first cluster core to build the first cluster. The method selects a second cluster core having the highest access frequency apart from edges of the first cluster, and assigns second edges in a second radius around second cluster core to build the second cluster. The method partitions the knowledge graph into a first sub-knowledge-graph comprising the first cluster and a second sub-knowledge-graph comprising the second cluster.Type: GrantFiled: November 27, 2018Date of Patent: August 10, 2021Assignee: International Business Machines CorporationInventors: Tim Uwe Scheideler, Erik Rueger, Stefan Ravizza, Frederik Frank Flöther
-
Publication number: 20210158195Abstract: Aspects of the present invention disclose a method for verifying labels of records of a dataset. The records comprise sample data and a related label out of a plurality of labels. The method includes one or more processors dividing the dataset into a training dataset comprising records relating to a selected label and an inference dataset comprising records with sample data relating to the selected label and all other labels out of the plurality of labels. The method further includes dividing the training dataset into a plurality of learner training datasets that comprise at least one sample relating to the selected label. The method further includes training a plurality of label-specific few-shot learners with one of the learner training datasets. The method further includes performing inference by the plurality of trained label-specific few-shot learners on the inference dataset to generate a plurality of sets of predicted label output values.Type: ApplicationFiled: November 26, 2019Publication date: May 27, 2021Inventors: Andrea Giovannini, Georgios Chaloulos, Frederik Frank Flother, Patrick Lustenberger, David Mesterhazy, Stefan Ravizza, Eric Slottke