Abstract: The techniques disclosed herein enable systems to perform repeatable and iterative load testing and performance benchmarking for artificial intelligence models deployed in a cloud computing environment. This is achieved by utilizing load profiles and representative workloads generated based on the load profiles to evaluate an artificial intelligence model under various workload contexts. The representative workload is then executed by the artificial intelligence model utilizing available computing infrastructure. Performance metrics are extracted from the execution and analyzed to provide insight into various performance dynamics such as the relationship between latency and data throughput. In addition, load profiles and input datasets are dynamically adjusted to evaluate different scenarios and use cases enabling the system to automatically test the artificial intelligence model across diverse applications.

Abstract: Provided are computing systems, methods, and platforms that train query processing models, such as large language models, to perform query intent classification tasks by using retrieval augmentation and multi-stage distillation. Unlabeled training examples of queries may be obtained, and a set of the training examples may be augmented with additional feature annotations to generate augmented training examples. A first query processing model may annotate the retrieval augmented queries to generate inferred labels for the augmented training examples. A second query processing model may be trained on the inferred labels, distilling the query processing model that was trained with retrieval augmentation into a non-retrieval augmented query processing model. The second query processing model may annotate the entire set of unlabeled training examples. Another stage of distillation may train a third query processing model using the entire set of unlabeled training examples without retrieval augmentation.

Abstract: Mechanisms are provided for compartmental epidemiological computer modeling based on mobility data. Machine learning training of an isolation rate prediction computer model is performed to generate a trained isolation rate prediction model that predicts an isolation rate of a biological population. Isolation data is received which comprises data indicating a measure of mobility of the biological population. The trained isolation rate prediction model is executed on input features extracted from the isolation data to generate a predicted isolation rate. A compartmental epidemiological computer model, comprising a plurality of compartments representing states of a population with regard to an infectious disease, is executed to simulate a progression of the infectious disease and flows of portions of the population from between compartments in the compartmental epidemiological computer model are controlled based on the predicted isolation rate.

Abstract: In some implementations, a probe tip assembly includes a driver printed circuit board assembly (PCBA) and a probe tip subassembly. The probe tip subassembly includes a plurality of probe tips, wherein a probe tip, of the plurality of probe tips, extends beyond an end of the PCBA, and the PCBA and the probe tip are configured to transmit an electric signal to test an optical component. The probe tip may include a material comprising at least one of copper (Cu), a beryllium copper (BeCu) alloy, tungsten (W), Paliney, Neyoro, and/or another conductive material.

Abstract: Disclosed are systems and techniques for performing mapping using radio frequency (RF) sensing. For instance, a server can obtain a first set of RF sensing data and orientation data corresponding to a first wireless device from a plurality of wireless devices. The first set of RF sensing data can be associated with at least one received waveform that is a reflection of a transmitted waveform from a first reflector. Based on the first set of RF sensing data, orientation data, and location data corresponding to the first wireless device, an indoor map can be generated that includes a reference to the reflector.

Abstract: Disclosed are systems and techniques for performing mapping using radio frequency (RF) sensing. For instance, a server can obtain a first set of RF sensing data and orientation data corresponding to a first wireless device from a plurality of wireless devices. The first set of RF sensing data can be associated with at least one received waveform that is a reflection of a transmitted waveform from a first reflector. Based on the first set of RF sensing data, orientation data, and location data corresponding to the first wireless device, an indoor map can be generated that includes a reference to the reflector.

Abstract: A medical episode analysis engine is provided. The engine generates a first matrix data structure having an entry for each concept pairing and storing a value representing relatedness weighted according to a temporal weighting function. The engine generates a second matrix data structure by calculating, for each entry in the first matrix, a relatedness measure of the concepts in the concept pairing based on a frequency of occurrence together. The engine generates, for each first concept, a concept embedding, based on the second matrix, that specifies, for each other second concept, a temporally weighted relatedness measure. The engine generates, for each anchor concepts, a corresponding episode definition comprising a plurality of related concepts corresponding to a same episode, based on the concept embedding. The engine processes new input data based on the episode definition data structures to identify instances of corresponding episodes in the new input data.

Abstract: In an approach for building a machine learning model with a flexible prediction horizon, a processor gathers statistical data related to a disease from one or more regional sources. A processor clusters the statistical data according to a plurality of localized regional source similarity criteria and a plurality of region criteria. A processor builds a plurality of training models based on the clustered statistical data. A processor builds a plurality of feature vectors based on the plurality of localized regional source similarity criteria and the plurality of region criteria. A processor trains the plurality of training models separately against the plurality of feature vectors. A processor selects a best performing training model for each of the plurality of localized regional source similarity criteria and the plurality of region criteria based on a performance criterion. A processor tests the best performing training model to predict one or more future outcomes.

Abstract: Mechanisms are provided to adapt computer modeling of an infectious disease based on noisy data and perform hyperlocal prediction of infectious disease dynamics and risks. Case report data is received and a trained background noise computer model is applied to generate first prediction results predicting infectious disease dynamics. The trained background noise computer model is trained to model infectious disease dynamics assuming that there is no community spread of the infectious disease. A first error measure of the first prediction results is determined and, in response to the first error measure being lower than a threshold value, the first prediction results are selected to output as predicted infectious disease dynamics. In response to the first error measure being equal/greater than the threshold value, second prediction results are selected. The second prediction results are generated by applying a trained infectious disease computer model to the received case report data.

Abstract: Mechanisms are provided to perform automatic case intervention detection in infectious disease case reports and for configuring an infectious disease computer model based on the automatic intervention detection. Case report data is received and a time ordered curve of the case report data is generated. One or more inflection points in the time ordered curve are identified. The one or more inflection points in the time ordered curve are correlated with one or more intervention entries specified in time stamped infectious disease intervention data, the one or more intervention entries specifying interventions implemented by authorities to control spread of the infectious disease. One or more model parameters of an infectious disease computer model are configured based on results of correlating the one or more inflection points with the one or more intervention entries.

Abstract: Mechanisms are provided to hypothetical scenario evaluations with regard to infectious disease dynamics based on similar regions. A user definition of a hypothetical scenario for a target region is received which specifies scenario features affecting an infectious disease spread amongst a population within the target region. Other predefined regions, in the set of predefined regions, having similar region characteristics to the target region are identified and attributes of the other predefined regions corresponding to the scenario features are identified. Modified model parameter(s) for an infectious disease computer model are derived based on the identified attributes. An instance of the infectious disease computer model is configured with the modified model parameter(s) and the instance is executed on case report data for the target region to generate a prediction for an infectious disease spread in the target region according to the hypothetical scenario, which is then output.

Abstract: Mechanisms are provided for hyperlocal prediction of epidemic dynamics and risks. Regional machine learning training is performed on an infectious disease computer model at least by: receiving first case report data; pre-processing the first case report data to remove noise at least by applying a smoothening algorithm to form first smoothed data; aggregating the first smoothed data into regional data, wherein aggregating the first smoothed data comprises correlating the first smoothed data to a target region corresponding to a population; and training the model using the regional data. The trained model is executed on new second case report data for the target region and automatic monitoring of performance of the model is performed according to a prediction accuracy of the model. In response to the prediction accuracy being below a predetermined threshold, automatic retraining is initiated.

Abstract: Mechanisms are provided for performing automated monitoring and retraining of infectious disease computer models. A trained infectious disease computer model is executed on case report data for a target region to generate prediction results predicting a state of an infectious disease spread within the target region for a given time. The prediction results generated by the trained infectious disease computer model are automatically compared to ground truth data to determine a deviation between the prediction results and the ground truth data. The ground truth data comprises at least one of actual case report data collected and reported by source computing systems for the given time, or a previous prediction result generated by the trained infectious disease computer model. Statistical test(s) are applied to the deviation to determine if it is statistically significant, and if so, re-training of the trained infectious disease computer model is automatically initiated.

Abstract: Mechanisms are provided for compartmental epidemiological computer modeling based on mobility data. Machine learning training of an isolation rate prediction computer model is performed to generate a trained isolation rate prediction model that predicts an isolation rate of a biological population. Isolation data is received which comprises data indicating a measure of mobility of the biological population. The trained isolation rate prediction model is executed on input features extracted from the isolation data to generate a predicted isolation rate. A compartmental epidemiological computer model, comprising a plurality of compartments representing states of a population with regard to an infectious disease, is executed to simulate a progression of the infectious disease and flows of portions of the population from between compartments in the compartmental epidemiological computer model are controlled based on the predicted isolation rate.

Abstract: A mechanism is provided in a data processing system to implement a model pipeline for predicting changes in disease transmission rate using a spatial temporal epidemiological model. The mechanism receives input data comprising disease case data for a disease and mobility data and prepares the input data to generate a training dataset, a validation dataset, and a test dataset. A feature selection module performs feature selection on the input data to select a first set of features for a binary classification computer model, a second set of features for a three-level classification computer model, and a third set of features for a regression computer model.

Abstract: Disclosed are systems and techniques for performing mapping using radio frequency (RF) sensing. For instance, a server can obtain a first set of RF sensing data and orientation data corresponding to a first wireless device from a plurality of wireless devices. The first set of RF sensing data can be associated with at least one received waveform that is a reflection of a transmitted waveform from a first reflector. Based on the first set of RF sensing data, orientation data, and location data corresponding to the first wireless device, an indoor map can be generated that includes a reference to the reflector.

Abstract: Mechanisms are provided for modifying a patient care plan or care provider workflow based on a patient risk assessment. The mechanisms analyze a patient medical record in a patient registry to identify at least one clinical measure for a corresponding patient and calculate a risk assessment value based on the at least one clinical measure value. The risk assessment value indicates a risk level for development of a medical condition or the occurrence of a medical event. The mechanisms select at least one of an action item or work flow to be performed to mitigate the risk level indicated by the risk assessment value based on the risk assessment value and a category of the risk assessment value. The mechanisms perform one or more operations for causing the action item to be performed or for performing the work flow.

Abstract: In some implementations, a probe tip assembly includes a driver printed circuit board assembly (PCBA) and a probe tip subassembly. The probe tip subassembly includes a plurality of probe tips, wherein a probe tip, of the plurality of probe tips, extends beyond an end of the PCBA, and the PCBA and the probe tip are configured to transmit an electric signal to test an optical component. The probe tip may include a material comprising at least one of copper (Cu), a beryllium copper (BeCu) alloy, tungsten (W), Paliney, Neyoro, and/or another conductive material.

Abstract: An optical device may include a substrate including a conductive core, a first layer stack on a first surface of the conductive core, a conductor-filled trench extending through the first layer stack to the conductive core such that the conductor-filled trench is on the first surface of the conductive core, and a second layer stack on a second surface of the conductive core. The optical device may include a vertical-cavity surface-emitting laser (VCSEL) chip above the conductor-filled trench. The VCSEL chip may include an array of VCSELs. A size of the conductor-filled trench may match a size of the VCSEL chip, match a size of an emission region of the array of VCSELs, or be greater than the size of the emission region of the array of VCSELs and less than the size of the VCSEL chip.

Abstract: In some implementations, a housing for an electro-optical device comprises a molded dielectric structural component, an electromagnetic interference (EMI) shield, and a plurality of conductive traces. The molded dielectric structural component may be configured to separate the EMI shield and the plurality of conductive traces.