Abstract: In one embodiment, a method herein comprises: inputting, by a device, an input prompt to a first large language model to generate an output; computing, by the device, a reward metric in part by using a solver to process the output; tuning, by the device and based on the reward metric, a second large language model configured to correct errors of the first large language model using reinforcement learning; and using, by the device, the second large language model to correct an error of the first large language model.

Abstract: In one implementation, a method is disclosed comprising: identifying, by a device, a plurality of functions within a source code based on one or more programmatic annotations of each of the plurality of functions within the source code; monitoring, by the device, execution characteristics associated with each of the plurality of functions within the source code during execution; constructing, by the device, a function call graph from the plurality of functions wherein each particular function in the function call graph is annotated with corresponding execution characteristics; and partitioning, by the device and based on the function call graph and one or more deployment specifications, the plurality of functions within the source code into singularly executable function capsules that meet the one or more deployment specifications.

Abstract: In one implementation, a method is disclosed comprising: associating, by a device in a service mesh, a security function with a portion of an online application that is executed in a distributed manner across the service mesh; executing, by the device, the security function and the portion of the online application within a trusted execution environment of the device to produce output data; generating, by the device, a cryptographic proof for the output data based on the security function; and providing, by the device, the output data and the cryptographic proof to a remote execution environment within the service mesh to establish a verifiable data lineage for the output data.

Abstract: In one embodiment, a device executing a first portion of a distributed application extracts label information from sensor data sent to the device by a sensor that indicates the sensor as a source of the sensor data and one or more data governance policies applicable to the sensor data. The device performs, based on the label information, a first data transformation of the distributed application on the sensor data using stored data, to form transformed data. The device forms combined label information for the transformed data by appending the label information with additional label information associated with the stored data and adding an indication of the first data transformation. The device provides the transformed data and combined label information to a remote device executing another portion of the distributed application.

Abstract: In one embodiment, a sidecar proxy executed by a device extracts label information from input data for input to a microservice associated with the sidecar proxy indicative of a lineage of the input data. The sidecar proxy makes, based on the label information, a determination as to whether the microservice processing the input data would violate a data governance policy. The sidecar proxy provides, based on the determination, the input data to the microservice. The sidecar proxy tags output data generated by the microservice with appended label information that includes the label information extracted from the input data and an indication of a data transformation performed by the microservice to the input data to form the output data.

Abstract: In one implementation, a device intercepts return data for an application programming interface call to be sent to a requester via a network. The device converts the return data into an embedding. The device determines a similarity between the embedding and one or more embeddings in a database that were generated from one or more documents deemed sensitive. The device blocks, based on the similarity, the return data from being sent via the network to the requester.

Abstract: In one implementation, a device generates outputs of nodes in a upstream layer of a partitioned neural network. The device assigns priorities to each of the outputs of the nodes. The device selects, based on the priorities, a subset of the outputs to send to a remote device. The device sends, via a computer network, the subset of the outputs to the remote device for input to a downstream layer of the partitioned neural network.

Abstract: In one implementation, a controller determines performance of a partitioned neural network. The controller identifies, based on the performance, a particular partition of the partitioned neural network as a bottleneck. The controller configures a first device to execute a replica of the particular partition. The controller configures a multiplexer that provides an output of the particular partition or the replica of the particular partition as input to a downstream partition of the partitioned neural network.

Abstract: In one implementation, a device receives a request to generate a set of video clips that depict a specified classification label. The device represents each of one or more objects depicted in a particular video clip over time as a set of timeseries of key points associated with that object. The device makes a determination as to whether the particular video clip depicts the specified classification label by analyzing the set of timeseries of key points associated with the particular video clip and in accordance with one or more constraint parameters. The device labels, based on the determination, the particular video clip with the specified classification label for inclusion in the set of video clips that depict the specified classification label.

Abstract: In one implementation, a device receives, via a user interface, one or more constraint parameters for each of a plurality of machine learning models that perform different analytics tasks. The device computes, based on the one or more constraint parameters, a set of weights for the plurality of machine learning models. The device generates a unified model by performing knowledge distillation on the plurality of machine learning models using the set of weights. The device deploys the unified model for execution by a particular node in a network.

Abstract: In one implementation, a device receives relevancy parameters from a user interface indicative of how relevant different portions of a text document are to a user. The device segments the text document into segments based on the relevancy parameters. The device generates a summary of the text document using the segments. The device provides, to the user interface, the summary of the text document and an indication of how much each of the segments contributed to the summary.

Abstract: In one embodiment, a first device in a network receives a quantum computing power metric indicative of a maximum available compute power of quantum computers. The first device receives, from a second device in the network, a listing of cryptographic suites available on the second device. The first device selects, based on the quantum computing power metric, a particular cryptographic suite from among the listing of cryptographic suites available on the second device. The first device sends, to the second device via the network, an indication that the particular cryptographic suite is to be used to encrypt and decrypt traffic exchanged between the first device and the second device.

Abstract: In one implementation, a device obtains a base machine learning model trained to label input data using a plurality of classes. The device receives a deployment task from a user interface indicative of a subset of one or more of the plurality of classes to be identified by a new model for deployment. The device selects a quantization level based on a difficulty associated with the deployment task. The device generates the new model for deployment that is quantized from the base machine learning model and specialized to label its input data using only the subset of one or more of the plurality of classes.

Abstract: A method to identify and remediate unsafe cipher suite deployments. The method includes identifying a target to be analyzed, determining a collection of cipher suites used by the target, generating and displaying a cipher suite dependency graph for the target based on the collection of cipher suites, and classifying the target as being one of post-quantum computing safe and post-quantum computing unsafe based on the cipher suite dependency graph and based on a predetermined rule for each cipher suite in the collection of cipher suites. A cipher suite hosted by a target can be remediated to convert the target to a post-quantum computing safe state.

Abstract: In one embodiment, a supervisory device in a federated learning system generates an aggregated model that aggregates a plurality of machine learning models trained by trainer nodes in a federated learning system during a training round. The supervisory device computes an accuracy loss metric for the aggregated model. The supervisory device also computes a fairness loss metric for the aggregated model based on fairness-related metrics associated with the plurality of machine learning models trained by the trainer nodes. The supervisory device initiates an additional training round during which the trainer nodes retrain their machine learning models for aggregation by the apparatus, in accordance with a constrained optimization problem that seeks to optimize a tradeoff between accuracy and fairness associated with the aggregated model.

Abstract: In one embodiment, a device identifies a plurality of tasks that a base machine learning model is able to perform. The device receives, via a user interface, a request to generate a specialized model to perform a particular task for deployment to a target deployment environment. The device uses knowledge distillation on the base machine learning model to train the specialized model to perform the particular task based on at least one of the plurality of tasks. The device causes the specialized model to be deployed to the target deployment environment.

Abstract: In one embodiment, a device makes a determination that performance of a global model generated by aggregating local models trained by a plurality of trainer nodes in a federated learning system has experienced a degradation. The device selects, in response to the determination, a particular trainer node from among the plurality of trainer nodes to generate debugging metrics. The device provides an indication that the particular trainer node is a root cause of the degradation.

Abstract: According to one or more embodiments of the disclosure, an example process herein may comprise: obtaining observability data for a computer system for a given time period; determining observability entities from the observability data; converting the observability entities into contextual vertices having associated vertex attributes; determining relationships among the contextual vertices based on correlation of the observability data; selecting a subset of the relationships to be edges based on a quality of the relationships, the edges having associated edge attributes; and generating an observability graph for the observability data for the computer system for the given time period by connecting the contextual vertices via corresponding edges.

Abstract: In one embodiment, a device represents each of a plurality of objects depicted in video data captured by a plurality of cameras over time as a set of key points associated with that object. The device forms, for each of the plurality of objects, a set of timeseries of the set of key points associated with that object. The device performs reidentification of a particular one of the plurality of objects across video data captured by two or more of the plurality of cameras by matching sets of timeseries of key points associated with that object derived from video data captured by two or more of the plurality of cameras. The device provides an indication of the reidentification for display to a user.

Abstract: In one embodiment, a device maintains a metamodel that describes a monitored system. The metamodel comprises a plurality of layers ranging from a sub-symbolic space to a symbolic space. The device tracks updates to the metamodel over time. The device updates the metamodel based in part on sub-symbolic time series data generated by the monitored system. The device receives, from a learning agent, a request for the updates to a particular layer of the metamodel associated with a specified time period. The device provides, to the learning agent, data indicative of one or more updates to the particular layer of the metamodel associated with the specified time period.