Abstract: The present disclosure is directed to systems and methods for identifying demographic information in a data file. The method may include: receiving the data file containing a plurality of fields of demographic information from a third-party, the data file having inconsistent or mislabeled nomenclatures for one or more fields of the plurality of fields or spurious demographic information; analyzing the data file using a machine learning model trained according to other data files to distinguish between each of the plurality of fields of demographic information, the machine learning model being based on a plurality of machine learning algorithms to identify different types demographic information; generating a score indicating a probability that each of the plurality of fields of demographic information was identified correctly; and generating a revised data file labeling each of the plurality of fields of demographic information based on the identified type.

Abstract: The present disclosure relates to the field of intelligent driving, and discloses an information fusion method. The method is applied to intelligent vehicles, and includes: obtaining a plurality of sensing information sets, where each sensing information set includes information about at least one target, and different sensing information sets come from sensing systems of different devices; and when at least two sensing information sets in the plurality of sensing information sets include information about a first target, fuse the information about the first target in the at least two sensing information sets to obtain fused information about the first target. In the method, information about a same target in a plurality of sources is fused to make the fused information about the target more accurate and effective.

Abstract: Disclosed is a method of artificial intelligence-based robotic pipetting for cell preparation. The method includes training an artificial intelligence/machine learning system (AI/ML system) to classify images of biological specimens. The method includes storing the classified images. The method includes receiving an image object from an imaging system that includes a microscopy system, a camera system configured to receive imaging from the microscopy system, and a lighting system configured to illuminate a biological material. The method includes processing the image object, using the AJ/ML system to determine a presence of a retrievable target cell. The method includes positioning a robotic pipettor at a target physical orientation relative to the retrievable target cell. The method includes confirming the robotic pipettor's location at the target physical orientation, using AI/ML system.

Abstract: An embodiment of a summarization application divides collected conversation data into media and text components. The application implements respective machine learning mechanisms to enhance modeling operations of the text and media components to identify key elements from the conversation. The application generates a headline banner from a group of key elements based on an analysis involving first predetermined criteria. The application also combines additional key elements to the group of key elements to form a second group of key elements. The application generates a summary from the second group of key elements based on a second analysis involving second predetermined criteria. The application presents, via a display, the headline banner according to a first output of the first key element analysis and the summary according to a second output of the second key element analysis.

Abstract: Embodiments described herein provide a method for text summarization. The method includes receiving a training dataset having at least an uncompressed text, a compressed text, and one or more information entities accompanying the compressed text. The method also includes generating, using a perturber model, a perturbed text with the one or more information entities being inserted into the compressed text. The method further includes training the perturber model based on a first training objective, and generating, using the trained perturber model, a perturbed summary in response to an input of a reference summary. The method further includes generating, via an editor model, a predicted summary by removing information from the perturbed summary conditioned on a source document of the reference summary, and training the editor model based on a second training objective.

Abstract: A method of explaining a machine learning model, including: receiving a plurality of disease states for a patient over time from a database, wherein the disease states have a plurality of features; generating a plurality of locally faithful explanation models for the patient for the disease states based upon the machine learning model; calculating an explanation with respect to one feature of the plurality of features over time using the locally faithful explanation models; and calculating the importance of the one feature of the plurality of features over time based upon the plurality of locally faithful explanation models.

Abstract: A system for constitutional analysis using objective functions includes a computing device configured to generate a ranked list of diseases, by determining a plurality of disease impact score vectors for plurality of diseases, and generating and optimizing a first objective function of the impact score vectors, to receive, from a user, a plurality of user physiological history data, to identify, as a function of a disease state classifier, a plurality of disease states associated with the plurality of user physiological history data, to match at least a disease state of the plurality of disease states to the ranked list of diseases, and to generate a curative habitual pattern to alleviate the at least a disease state by combining intervention elements to form a curative habitual pattern candidates, calculating a curative impact score of each curative habitual pattern candidate, and selecting the curative habitual pattern using the curative impact score.

Abstract: A computing apparatus comprises one or more computer readable storage media, one or more processors operatively coupled with the one or more computer readable storage media, and program instructions stored on the one or more computer readable storage media. The program instructions, when executed by the one or more processors, direct the computing apparatus to at least generate an audio recording of speech, extract features from the audio recording indicative of vocal patterns in the speech, determine a register classification of the speech based at least on the features, and display an indication of the register classification in a user interface.

Abstract: A method and device estimate life consumption for an electronic component for use with an electric motor. The method and device allow maximal/minimal temperature values of semiconductor devices to be continuously monitored to allow on-cycle thermal damage and post-cycle damage to be determined in a parallel manner.

Abstract: A system and method for generating a radiological report. The system and method displaying, on a display, an image of a region of interest, determining an image characteristic of the region of interest, determining, via a processor, a recommendation based on the image characteristic, and generating, via the processor, a report including the recommendation.

Abstract: An apparatus for a precise analysis of a severity of arthritis includes an image collection unit configured to collect a medical image having captured a joint of a user, a region detection unit configured to detect one or more regions of interest for analyzing arthritis in the medical image through a learned automatic region detection model, an individual analysis unit configured to extract quantitative feature values from the detected regions of interest and derives one or more individual analysis data from among a severity of arthritis, a severity of osteoproliferation, and a severity of hardness of a subchondral bone based on the feature values, and an integrated analysis unit configured to finely classify a severity of degenerative arthritis through an integrated analysis model learned based on the individual analysis data.

Abstract: Disclosed are systems and methods for improving interactions with and between computers in content providing, searching and/or hosting systems supported by or configured with devices, servers and/or platforms. The disclosed systems and methods provide a novel framework for compiling, updating and dynamically managing a confidence graph for a user that leads to generation of a scored interest profile for the user that content providers can utilize as a basis for disseminating their proprietary digital content. The disclosed confidence graph provides a scored interest profile for each user that is based on authenticated user data derived from an inbox of the user. The confidence graph is not only derived from authenticated data, but is also dynamic and evolves simultaneously with changing user interests. Thus, digital content is selected and transmitted to users based on the current, real-time digital data reflecting their current interests as reflected by their inbox activity.

Abstract: A smart space may be provided by a hub and an artificial intelligence server in communication with the hub. The hub may receive data from at least one smart object in the smart space. The artificial intelligence server may generate clusters of the data received from each of the at least one smart objects. The server may perform processing comprising using a cluster to detect an anomaly in the smart object, identify the smart object, classify the smart object, determine a user behavior, determine a user mood, determine an energy consumption pattern, or create an automated action, or a combination thereof.

Abstract: This document relates to training of machine learning models. One example method involves providing a machine learning model having a first classification layer, a second classification layer, and an encoder that feeds into the first classification layer and the second classification layer. The example method also involves obtaining first training examples having explicit labels and second training examples having inferred labels. The inferred labels are based at least on actions associated with the second training examples. The example method also involves training the machine learning model using the first training examples and the second training examples using a training objective that considers first training loss of the first classification layer for the explicit labels and second training loss of the second classification layer for the inferred labels. The method also involves outputting a trained machine learning model having the encoder and the first classification layer.

Abstract: Disclosed are systems and methods utilizing neural contextual bandit for improving interactions with and between computers in content generating, searching, hosting and/or providing systems supported by or configured with personal computing devices, servers and/or platforms. The systems interact to make item recommendations using latent relations and latent representations, which can improve the quality of data used in processing interactions between or among processors in such systems. The disclosed systems and methods use neural network modeling in automatic selection of a number of items for recommendation to a user and using feedback in connection with the recommendation for further training of the model(s).

Abstract: An AI-based new learning model generation system for vision inspection on a product production line is proposed. In the AI-based new learning model generation system, the candidate set extraction module extracts two or more candidate data sets on the basis of determination type information from among a plurality of training data sets that have been applied to learning of existing learning models previously generated for the vision inspection on the product production line. In addition, an additional set determination module calculates similarity between training images of new training data and a candidate data set, and determines any one greater than or equal to a reference value as an additional training data. In addition, the new model generation module may generate a new learning model by training the additional training data set and the new training data as a pre-training model.

Abstract: Systems and methods for tuning hyperparameters of a model include receiving a tuning request for tuning hyperparameters, the tuning request includes a first and a second objective function for the machine learning model. The first and second objective functions may output metric values that do not improve uniformly. Systems and methods additionally include defining a joint tuning function that is based on a combination of the first and second objective functions; executing a tuning operation; identifying a Pareto efficient frontier curve defined by a plurality of distinct hyperparameter values; applying metric thresholds to the Pareto efficient frontier curve; demarcating the Pareto efficient frontier curve into at least a first infeasible section and a second feasible section; searching the second feasible section of the Pareto efficient frontier curve for one or more proposed hyperparameter values; and identifying at least a first set of proposed hyperparameter values based on the search.

Abstract: At least one request to store diagnostic state of a virtual machine is obtained. Based on obtaining the at least one request, a store of diagnostic state of the virtual machine is performed to provide stored diagnostic state of the virtual machine. The performing the store includes encrypting the diagnostic state of the virtual machine that is unencrypted and being stored to prevent a reading of the diagnostic state of the virtual machine by an untrusted entity prior to encrypting the diagnostic state of the virtual machine that is unencrypted and being stored.

Abstract: A system and method for assisting entities or users to make decisions is disclosed. The user access the decision-making engine (MG Case composite) for end-to-end decision-making process. The system comprises a computing device (physical or on the cloud) having a processor and a memory, and a database. The engine comprises multiple modules such as case open, structure, MG Case matrix, brainstorming, problem solving, data analysis, speed math, and case end along with drive and alignment, integration and transition, and insights and impacts. The system is an integrated system and the modules are executed by the processor to perform an operation that draws on modules as needed. The decision-making engine interacts with internal and external facets of an organization, including users, various communication systems, ERPs, databases, etc. to execute various tasks related to decision making such as gathering data and driving the decision-making process through the organization or with the user.

Abstract: A method for automatically generating news event of a certain topic applied in an electronic device analyzes text of the news event by a topic model to obtain topics, a probability distribution of keywords in each topic is established, and a time interval distribution of the keywords in each topic is calculated. Keywords within a preset probability distribution range are selected to reduce the size of a word bag relating to the topic, and a time interval range of the reduced word bag of the topic is determined. A calculation of text similarities of the text in a database is made to obtain a news article corresponding to each topic according to the time interval range of the reduced word bag, and a title of the news article as a target topic of the text of the news event is determined.

Abstract: A method of workload management for distributing workload over a single instruction and multiple data (SIMD) based parallel processing architecture includes pre-processing of vert large ontology file on the host processor. The outcome of the preprocessing step is a set of arrays which are loaded on the SIMD based parallel processing architecture to process the ontology file over input text documents of any kind and generate the tagged outcome. The method provides the maximum granularity to facilitate allocation of the maximum number of software threads on the SIMD based parallel processing architecture to achieve minimum document processing latency.

Abstract: A device receives first transaction information associated with a first transaction, and a first transaction account utilized for the first transaction and associated with a first financial institution. The device determines, based on a fraud model, that the first transaction is to be denied due to potential fraud associated with the first transaction account and receives second transaction information associated with a second transaction, and a second transaction account utilized for the second transaction and associated with a second financial institution. The device processes the first transaction information and the second transaction information, with a matching model, to determine whether the first transaction information matches the second transaction information and determines that the first transaction was incorrectly denied when the first transaction information matches the second transaction information within a predetermined threshold.

Abstract: A system determines baseline deployment properties of operating system deployments stored by a deployment repository and endpoint deployment properties of a deployed operating system executed by an endpoint device. An artificial intelligence model is configured to determine a security response based at least in part on the endpoint deployment properties of the endpoint device. By providing the endpoint deployment properties to the artificial intelligence model, a mismatch value is determined that corresponds to an amount that the endpoint deployment properties are different than the baseline deployment properties. Based on the mismatch value, an action is determined to improve security of the deployed operating system executed by the endpoint device. The determined action is executed to improve security of the deployed operating system.

Abstract: Methods and systems are provided for training a neural network with augmented data. A dataset comprising a plurality of classes is obtained for training a neural network. Prior to initiation of training, the dataset may be augmented by performing affine transformations of the data in the dataset, wherein the amount of augmentation is determined by a data augmentation variable. The neural network is trained with the augmented dataset. A training loss and a difference of class accuracy for each class is determined. The data augmentation variable is updated based on the total loss and class accuracy for each class. The dataset is augmented by performing affine transformations of the data in the dataset according to the updated data augmentation variable, and the neural network is trained with the augmented dataset.

Abstract: Example techniques are provided for the task of differentially private clustering. For several basic clustering problems, including Euclidean DensestBall, 1-Cluster, k-means, and k-median, the present disclosure provides efficient differentially private algorithms that achieve essentially the same approximation ratios as those that can be obtained by any non-private algorithm, while incurring only small additive errors. This improves upon existing efficient algorithms that only achieve some large constant approximation factors.

Abstract: An approach for selecting a transformed prediction model for estimating the uncertainty of prediction associated with machine learning is disclosed. The approach determines a function corresponding to a prediction interval based on one or more datasets and calculates one or more sets of prediction interval parameters associated with the function based on training a set of regression models with the one or more datasets. The approach creates one or more transformed predictions based on the one or more sets of parameters and based on a logical distance and selects a transformed prediction model based on a mean absolute correlation from the one or more transformed predictions. Furthermore, the approach outputs the selected transformed prediction model.

Abstract: Machine-learning-aided automatic taxonomy for web data. In an embodiment, a training dataset of annotated features is used to train a model to predict a class in a taxonomy of web-based activities. The features may be derived from a uniform resource locator (URL) of an online resource and associated metadata. During operation, the features may be extracted from the URL and metadata of each activity record in web data. The trained model may be applied to the extracted features for each activity record to predict a class within the taxonomy. The predicted taxonomic class may be stored in association with the URL that was extracted from the activity record to produce a taxonomized URL.

Abstract: An anomalous behavior detector has been designed to detect novel behavioral changes of devices based on network traffic data that likely correlate to anomalous behaviors. The anomalous behavior detector uses the local outlier factor (LOF) algorithm with novelty detection. After initial semi-supervised training with a single class training dataset representing stable device behaviors, the obtained model continues learning frontiers that delimit subspaces of inlier observations with live network traffic data. Instead of traffic variables being used as features, the features that form feature vectors are similarities of network traffic variable values across time intervals. A feature vector for the anomalous behavior detector represents stability or similarity of network traffic variables that have been chosen as device identifiers and behavioral indicators.

Abstract: A computer-implemented method may include receiving at least three video clips of a sporting event, where each of the video clips may (i) be simultaneously captured over at least a portion of time, and (ii) include at least one common player wearing an indicia on a jersey that is distinguishing from indicia on other players. Tracking locations of the at least one common player captured in the at least three video clips may be generated by triangulating distances of the common player(s) in the video clips. Statistical information of the common player(s) may be generated from the tracking locations. The common player(s) may be represented on a graphical display. The common player(s) may be controlled by applying at least one of the tracking locations and statistical information of the common player(s).

Abstract: This document relates to training of machine learning models. One example method involves providing a machine learning model having a first classification layer, a second classification layer, and an encoder that feeds into the first classification layer and the second classification layer. The example method also involves obtaining first training examples having explicit labels and second training examples having inferred labels. The inferred labels are based at least on actions associated with the second training examples. The example method also involves training the machine learning model using the first training examples and the second training examples using a training objective that considers first training loss of the first classification layer for the explicit labels and second training loss of the second classification layer for the inferred labels. The method also involves outputting a trained machine learning model having the encoder and the first classification layer.

Abstract: A method for identifying from among network-connected devices a particular device likely associated with a theft of distributed content includes obtaining content, the content having been distributed from a particular one of the network-connected devices, identifying in the obtained content anti-collusion codes, and determining the particular one of the network-connected devices is likely associated with the theft of distributed content when an aggregated probability calculated using the identified anti-collusion codes is equal to or exceeds a predetermined threshold.

Abstract: A customized highlight sequence depicting multiple events, and based on a user's personal characteristics, interests, and/or preferences, is generated and presented. The events can be sporting events, entertainment events, and/or the like. For example, in the context of sporting events, a determination is made as to what types of sports, teams, leagues, players, plays, and/or the like are of interest to the user. In at least one embodiment, the amount of time available to the user can be obtained, so that the length of the highlight sequence can be tailored to the available time. A customized highlight sequence is then generated and presented, containing those specific portions of the sporting events that are likely to be of interest, arranged in a manner that is likely to be entertaining to the user and comports with the time restrictions.

Abstract: Systems and methods are provided for implementing a tool for evaluating an effect on an event, such as a medication or treatment, on a subject's condition, using a propensity model that identifies matched treatment and control cohorts within a base population of subjects. A propensity value threshold, which can be obtained based on user input, can be used to adjust the selection of subjects for treatment and control cohorts. The tool allows analyzing features of the subjects in the treatment and control groups, and further allows for evaluation and comparison of survival objectives of subjects in the treatment and control groups.

Abstract: A machine learning system builds and uses control policies for controlling robotic performance of a task. Such control policies may be trained using targeted updates. For example, two trials identified as similar may be compared and evaluated to determine which trial achieved a greater degree of task success; a control policy update may then be generated based on identified differences between the two trials.

Abstract: Systems and methods for discovering and annotating geo-fences from geo-referenced data are disclosed. The systems and methods input an area of interest containing a plurality of geo-referenced points having associated labels, and divides the area interest into cells. Each cell is assigned an initial label from among the plurality of labels and hierarchical clustering is used to find clusters of cells having a common label based on a maximization of an objective function for each cell with the objective function being dependent upon favoring spatially adjacent cells having a common label and limiting overgeneralization of the common label.

Abstract: A method of explaining a machine learning model, including: receiving a plurality of disease states for a patient over time from a database, wherein the disease states have a plurality of features; generating a plurality of locally faithful explanation models for the patient for each disease state based upon the machine learning model; calculating an explanation with respect to one feature of the plurality of features over time using the locally faithful explanation models; and calculating the importance of the one feature of the plurality of features over time based upon the plurality of locally faithful explanation models.

Abstract: Provided herein are biomarkers and methods for generating scores useful for assessing psoriatic arthritis (PsA) disease activity in subjects previously diagnosed with PsA. The invention also provides predictive model methods based on the biomarkers, as well as computer systems, software embodiments of the models for scoring and optionally classifying samples, and methods of recommending optimal therapeutic regimens.

Abstract: A method for identifying and ranking potentially privileged documents using a machine learning topic model may include receiving a set of documents. The method may also include, for each of two or more documents in the set of documents, extracting a set of spans from the document, generating, using a machine learning topic model, a set of topics and a subset of legal topics for the set of spans, generating a vector of probabilities for each span with a probability being assigned to each topic in the set of topics for the span, assigning a score to one or more spans in the set of spans by summing the probabilities in the vector that are assigned to a topic in the subset of legal topics, and assigning a score to the document. The method may further include ranking the two or more documents by their assigned scores.

Abstract: Computer-implemented systems and methods are provided for constructing a numerical model to generate a probability that a patient diagnosed with multiple myeloma will die within a predetermined period of time. Reference data for a plurality of patients diagnosed with multiple myeloma is received. First variables selected from the reference data are deemed predictive of mortality for a first predetermined period of time, and second variables selected from the reference data are deemed predictive of mortality for a second predetermined period of time. A first computer model comprising a combination of variables of the first selected variables and first weighting factors is generated. A second computer model comprising a combination of variables of the second selected variables and second weighting factors is generated. The first computer model and the second computer model are trained using the reference data to determine numerical values for the respective first and second weighting factors.

Abstract: Systems and methods disclosed herein provide for a case-based reasoning using universal ontologies. Embodiments of the systems and methods provide for comparing current and past cases based on the universal ontologies and sorting the past cases based on the comparison, wherein the universal ontology integrates authority information associated with the current and past cases.

Abstract: A method for displaying points of interest on a digital map on a display, each point of interest being assigned at least one category and the space in front of the display being segmented into spatial regions where each spatial region is assigned a category. The method includes detecting the position of a hand of a user or part of the hand of the user by sensors, more particularly by cameras, identifying the spatial region in which the position of the hand or the part of the hand lies, identifying the category assigned to the identified spatial region, and displaying or highlighting the points of interest that have been assigned to the identified category.

Abstract: Methods, devices and system for determining fasting glucose level information and post-prandial glucose level information for diagnosing pre-diabetic and diabetic conditions based on monitored glucose measurements are provided.

Abstract: A medical system comprises processing circuitry configured to: receive a first trained model, wherein the trained model has been trained using a first data set acquired in a first cohort; receive a second data set acquired in a second cohort; input data included in the second data set and data representative of the first trained model into a second trained model; and receive from the second trained model an affinity-relating value which represents an affinity between the data included in the second data set and the first trained model.

Abstract: Aspects of the present disclosure enable humanly-specified relationships to contribute to a mapping that enables compression of the output structure of a machine-learned model. An exponential model such as a maximum entropy model can leverage a machine-learned embedding and the mapping to produce a classification output. In such fashion, the feature discovery capabilities of machine-learned models (e.g., deep networks) can be synergistically combined with relationships developed based on human understanding of the structural nature of the problem to be solved, thereby enabling compression of model output structures without significant loss of accuracy. These compressed models provide improved applicability to “on device” or other resource-constrained scenarios.

Abstract: In some embodiments, a method receives a connection to a data source. The method analyzes metadata of the data source to determine a first type of metadata for a first type of database access and a second type of metadata for a second type of database access. The first type of metadata and the second type of metadata are combined into a data structure. Then, the method stores the data structure where the data structure is used to analyze a query to determine which of the first type of database access and the second type of database access to use for the query.

Abstract: A physical environment is equipped with a plurality of sensors (e.g., motion sensors). As individuals perform various activities within the physical environment, sensor readings are received from one or more of the sensors. Based on the sensor readings, activities being performed by the individuals are recognized and the sensor data is labeled based on the recognized activities. Future activity occurrences are predicted based on the labeled sensor data. Activity prompts may be generated and/or facility automation may be performed for one or more future activity occurrences.

Abstract: A fault finding support system is a fault finding support system for instructing a maintenance worker of an appropriate examination location and examination contents using fault knowledge data in which a causal relationship of fault of a target machine is described, the system including: an examination procedure creation unit that creates an examination procedure with respect to the examination location and the examination contents of the target machine using the fault knowledge data, an examination result storage unit that stores an examination history of performing examination using the examination procedure created by the examination procedure creation unit, an update target extraction unit that recommends an update location of the fault knowledge data using the examination history stored in the examination result storage unit, and a user interface that provides a function of displaying a location extracted by the update target extraction unit and updating the fault knowledge data.

Abstract: A method for communicating information relating to test results of a user includes obtaining test results of a user. An assertion is derived from the test results of the user. The test results are input to a pre-provisioned first algorithm. The assertion is encapsulated in a first data object by a PGE that controls an environment in which the first algorithm is executed. A first proof is generated which is configured to be usable to verify that the first algorithm used the test results to produce the assertion when provided to a PVE along with the first data object. The test results itself are excluded from the first proof and the first data object such that privacy of the test results is maintained. The first proof and the first data object are communicated to a receiving communication device from an enterprise communication device associated with the user and an enterprise.

Abstract: Methods, systems, apparatus, and computer programs are disclosed for software-accelerated genomic data read mapping. In one aspect, the methods can include actions of obtaining a k-mer seed from a genomic data read, generating a genomic signature based on the obtained k-mer seed, determining a reference sequence location that match at least a portion of the k-mer seed using a hash data structure, wherein the hash data structure comprises N data cells comprising a first portion storing a predetermined genomic signature and a second portion storing a value that corresponds to a first occurrence of a reference sequence location that match at least a portion of the k-mer seed from which the predetermined genomic signature was derived, and selecting the determined reference sequence location as an actual alignment for the obtained k-mer seed based on one or more alignment scores.

Abstract: A method and system is provided for assisting a user to assign a label to words or spans of text in a transcript of a conversation between a patient and a medical professional and form groupings of such labelled words or spans of text in the transcript. The transcript is displayed on an interface of a workstation. A tool is provided for highlighting spans of text in the transcript consisting of one or more words. Another tool is provided for assigning a label to the highlighted spans of text. This tool includes a feature enabling searching through a set of predefined labels available for assignment to the highlighted span of text. The predefined labels encode medical entities and attributes of the medical entities. The interface further includes a tool for creating groupings of related highlighted spans of texts. The tools can consist of mouse action or key strokes or a combination thereof.