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 retrieves a set of documents based on their relevancy to an input query from a user interface. The device extracts excerpts of varying sizes from the set of documents that are relevant to the input query. The device performs a ranking of the excerpts based on their relevancy to the input query. The device augments, based on the ranking, the input query based on one or more of the excerpts to form a prompt for input to a language model.

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 embodiment, a method for a cache-generated frequently asked questions page includes converting a received query into a set of embeddings and performing a semantic searching operation based on information contained in the set of embeddings against a cache of question and answer pairs. The method further includes returning a stored answer to the received query responsive to a determination that a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level for the information contained in the set of embeddings, the stored answer derived from the particular question and answer pair and performing a large language model operation to generate an answer to the query responsive to a determination that no question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings.

Abstract: In one embodiment, a device identifies, using a semantic reasoning engine, activities in a location, based on sensor data obtained from a plurality of sensors deployed to the location. The device associates the activities with areas of the location in which they occurred. The device makes, using the semantic reasoning engine, an inference about a particular activity, based in part on where that activity occurred. The device raises, based on the inference, an alert regarding the particular activity.

Abstract: A device obtains a particular query for input to a language model. The device identifies a plurality of cached query-response pairs whose queries are similar to that of the particular query. The device uses a verification model to assign joint probabilities between the particular query and responses from the plurality of cached query-response pairs. The device, based on the joint probabilities provides a particular response from the plurality of cached query-response pairs, in lieu of using the particular query as input to the language model to generate a new response.

Abstract: In one implementation, a device receives, via a user interface, a selection of a labeled training dataset and a selection of an unlabeled training dataset, wherein the unlabeled training dataset is captured from a target domain. The device forms a domain-adapted training dataset by pruning the labeled training dataset based on the unlabeled training dataset. The device trains a machine learning model using the domain-adapted training dataset. The device prunes the machine learning model to form a domain-adapted model for the target domain.

Abstract: In one embodiment, a method for a cache-generated frequently asked questions page includes converting a received query into a set of embeddings and performing a semantic searching operation based on information contained in the set of embeddings against a cache of question and answer pairs. The method further includes returning a stored answer to the received query responsive to a determination that a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level for the information contained in the set of embeddings, the stored answer derived from the particular question and answer pair and performing a large language model operation to generate an answer to the query responsive to a determination that no question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings.

Abstract: In one embodiment, a device in a federated learning system receives a global model from an aggregation node. The device applies noise to the global model, to form a noise-augmented model. The device performs local training using the noise-augmented model and a local training dataset, to form a local model. The device provides, via a network, the local model to the aggregation node for aggregation with other local models trained in the federated learning system.

Abstract: In one embodiment, an illustrative process herein may comprise: accessing, by a device, a plurality of machine learning models; determining, by the device, one or more objective functions for the plurality of machine learning models; evaluating, by the device, the plurality of machine learning models against the one or more objective functions to establish a comparative assessment of each of the plurality of machine learning models for the one or more objective functions; and providing, by the device, the comparative assessment of each of the plurality of machine learning models to a model selection process for selection of a specific machine learning model from the plurality of machine learning models for a given objective function of the one or more objective functions.

Abstract: This disclosure describes techniques for protecting privacy of a user with respect to emotion detection via a computer network. The techniques may include receiving sensed data associated with a user. A privacy policy of the user may be used with processing of the sensed data. For example, based at least in part on the privacy policy, a private subset of the sensed data may be filtered from remaining sensed data. The remaining sensed data may be used to determine an emotion classification result. The emotion classification result may indicate a sharable emotion of the user, for instance.

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 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 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 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 controller for a federated learning system identifies a first dataset and a second dataset available to a particular node of the federated learning system. The first dataset comprises features that are common to all nodes of the federated learning system. The second dataset comprises features that are common only to a subset of nodes of the federated learning system. The controller configures the particular node to train a first model using the first dataset. The controller causes formation of a global model in the federated learning system that aggregates the first model from the particular node with models from all other nodes of the federated learning system. The controller configures the particular node to train a second model using the second dataset.

Abstract: In one embodiment, a device makes an inference regarding input data using an artificial intelligence model. The device captures one or more feature vectors used by the artificial intelligence model to make the inference. The device selects, based on the one or more feature vectors, a representative sample from a training dataset used to train the artificial intelligence model. The device provides the representative sample for display in conjunction with the inference.

Abstract: In one embodiment, a device identifies, using a semantic reasoning engine, activities in a location, based on sensor data obtained from a plurality of sensors deployed to the location. The device associates the activities with areas of the location in which they occurred. The device makes, using the semantic reasoning engine, an inference about a particular activity, based in part on where that activity occurred. The device raises, based on the inference, an alert regarding the particular activity.

Abstract: In one embodiment, a device receives, from a plurality of training nodes that train a set of machine learning models using local training datasets, bias metrics associated with those machine learning models for each feature of the local training datasets. The device generates aggregated machine learning models over time that aggregate the machine learning models trained by the plurality of training nodes. The device constructs, based on the bias metrics, bias lineages for the aggregated machine learning models. The device provides, based on the bias lineages, a bias lineage for a particular one of the aggregated machine learning models for display.