Abstract: Systems, methods, and computer-readable for multi-spatial scale object detection include generating one or more object trackers for tracking at least one object detected from on one or more images. One or more blobs are generated for the at least one object based on tracking motion associated with the at least one object. One or more tracklets are generated for the at least one object based on associating the one or more object trackers and the one or more blobs, the one or more tracklets including one or more scales of object tracking data for the at least one object. One or more uncertainty metrics are generated using the one or more object trackers and an embedding of the one or more tracklets. A training module for detecting and tracking the at least one object using the embedding and the one or more uncertainty metrics is generated using deep learning techniques.

Abstract: Systems, methods, and computer-readable for multi-temporal scale analysis include obtaining two or more timescales associated with one or more images. A context associated with a monitoring objective is obtained, based on real time analytics or domain specific knowledge. The monitoring objective can include object detection, event detection, pattern recognition, or other. At least a subset of timescales for performing a differential analysis on the one or more images is determined based on the context. Multi timescale surprise detection and clustering are performed using the subset of timescales to determine whether any alerts are to be generated based on entropy based surprises. A set of rules can be created for the monitoring objective based on the differential analytics and alerts or entropy based surprises, if any.

Abstract: Systems, methods, and computer-readable for multi-spatial scale object detection include generating one or more object trackers for tracking at least one object detected from on one or more images. One or more blobs are generated for the at least one object based on tracking motion associated with the at least one object. One or more tracklets are generated for the at least one object based on associating the one or more object trackers and the one or more blobs, the one or more tracklets including one or more scales of object tracking data for the at least one object. One or more uncertainty metrics are generated using the one or more object trackers and an embedding of the one or more tracklets. A training module for detecting and tracking the at least one object using the embedding and the one or more uncertainty metrics is generated using deep learning techniques.

Abstract: Systems, methods, and computer-readable for multi-spatial scale object detection include generating one or more object trackers for tracking at least one object detected from on one or more images. One or more blobs are generated for the at least one object based on tracking motion associated with the at least one object. One or more tracklets are generated for the at least one object based on associating the one or more object trackers and the one or more blobs, the one or more tracklets including one or more scales of object tracking data for the at least one object. One or more uncertainty metrics are generated using the one or more object trackers and an embedding of the one or more tracklets. A training module for detecting and tracking the at least one object using the embedding and the one or more uncertainty metrics is generated using deep learning techniques.

Abstract: Systems and methods are disclosed for anticipating a video switch to accommodate a new speaker in a video conference comprising a real time video stream captured by a camera local to a first videoconference endpoint is analyzed according to at least one speaker anticipation model. The speaker anticipation model predicts that a new speaker is about to speak. Video of the anticipated new speaker is sent to the conferencing server in response to a request for the video on the anticipated new speaker from the conferencing server. Video of the anticipated new speaker is distributed to at least a second videoconference endpoint.

Abstract: Various implementations disclosed herein provide a search engine that receives a search request from a sensor gateway, and provides search results in return. In various implementations, the search request includes a first set of measurements captured by a first sensor, a first measurement from the first set of measurements is outside a defined range. In various implementations, the search engine determines a first feature vector based on the first set of measurements, and identifies a second feature vector that indicates a second set of measurements within a degree of similarity to the first set of measurements. In some implementations, the second set of measurements are captured by a second sensor. In various implementations, the search engine determines a search result based on the second feature vector, and transmits the search result. In some implementations, the search result indicates one or more instructions executable by the first sensor.

Abstract: Systems, methods, and computer-readable for multi-spatial scale object detection include generating one or more object trackers for tracking at least one object detected from on one or more images. One or more blobs are generated for the at least one object based on tracking motion associated with the at least one object. One or more tracklets are generated for the at least one object based on associating the one or more object trackers and the one or more blobs, the one or more tracklets including one or more scales of object tracking data for the at least one object. One or more uncertainty metrics are generated using the one or more object trackers and an embedding of the one or more tracklets. A training module for detecting and tracking the at least one object using the embedding and the one or more uncertainty metrics is generated using deep learning techniques.

Abstract: Systems, methods, and computer-readable for multi-temporal scale analysis include obtaining two or more timescales associated with one or more images. A context associated with a monitoring objective is obtained, based on real time analytics or domain specific knowledge. The monitoring objective can include object detection, event detection, pattern recognition, or other. At least a subset of timescales for performing a differential analysis on the one or more images is determined based on the context. Multi timescale surprise detection and clustering are performed using the subset of timescales to determine whether any alerts are to be generated based on entropy based surprises. A set of rules can be created for the monitoring objective based on the differential analytics and alerts or entropy based surprises, if any.

Abstract: In one embodiment, a method includes training a deep neural network using a first set of network characteristics corresponding to a first time and a second set of network characteristics corresponding to a second time, generating, using the deep neural network, a predictive set of network characteristics corresponding to a future time, and assigning a task of a distributed application to a processing unit based on the predictive set of network characteristics.

Abstract: In one embodiment, a method includes receiving current data, the current data including time series data representing a plurality of time instances. The method includes storing at least a recent portion of the current data in a buffer. The method includes reducing the dimensionality of the current data to generate dimensionality-reduced data. The method includes generating a reconstruction error based on the dimensionality-reduced data and a plurality of neural network metrics. At least one of a size of the recent portion of the current data stored in the buffer or an amount of the reducing the dimensionality of the current data is based on the reconstruction error.

Abstract: Systems and methods are disclosed for anticipating a video switch to accommodate a new speaker in a video conference comprising a real time video stream captured by a camera local to a first videoconference endpoint is analyzed according to at least one speaker anticipation model. The speaker anticipation model predicts that a new speaker is about to speak. Video of the anticipated new speaker is sent to the conferencing server in response to a request for the video on the anticipated new speaker from the conferencing server. Video of the anticipated new speaker is distributed to at least a second videoconference endpoint.

Abstract: In one illustrative example, a network cybersecurity procedure may be employed with use of at least one unmanned aerial vehicle (UAV), where the UAV includes an intermediary pairing device for providing a temporary connection between a first network (e.g. a private LAN) and a second network (e.g. the Internet). The network cybersecurity procedure may involve deploying the UAV in proximity to the first network, such that the intermediary pairing device pairs with a first pairing device via a first transceiver and with a second pairing device via a second transceiver. A temporary connection is established between the first network connected via the first pairing device and the second network connected via the second pairing device. Data is communicated between a first device (e.g. IoT device) or server of the first network and a second device or server of the second network over the temporary connection. During this time, the intermediary pairing device executes a cybersecurity service function.

Abstract: Systems and methods are disclosed for anticipating a video switch to accommodate a new speaker in a video conference comprising a real time video stream captured by a camera local to a first videoconference endpoint is analyzed according to at least one speaker anticipation model. The speaker anticipation model predicts that a new speaker is about to speak. Video of the anticipated new speaker is sent to the conferencing server in response to a request for the video on the anticipated new speaker from the conferencing server. Video of the anticipated new speaker is distributed to at least a second videoconference endpoint.

Abstract: In one embodiment, a method includes obtaining text from a user, applying the text to a deep learning neural network to generate a plurality of bias coordinates defining a point in an embedded space, and, in response to determining that at least one of the plurality of bias coordinates exceeds a threshold, providing an indication of bias to the user.

Abstract: In one illustrative example, a network cybersecurity procedure may be employed with use of at least one unmanned aerial vehicle (UAV), where the UAV includes an intermediary pairing device for providing a temporary connection between a first network (e.g. a private LAN) and a second network (e.g. the Internet). The network cybersecurity procedure may involve deploying the UAV in proximity to the first network, such that the intermediary pairing device pairs with a first pairing device via a first transceiver and with a second pairing device via a second transceiver. A temporary connection is established between the first network connected via the first pairing device and the second network connected via the second pairing device. Data is communicated between a first device (e.g. IoT device) or server of the first network and a second device or server of the second network over the temporary connection. During this time, the intermediary pairing device executes a cybersecurity service function.

Abstract: In one embodiment, a method includes receiving current data, the current data including time series data representing a plurality of time instances. The method includes storing at least a recent portion of the current data in a buffer. The method includes reducing the dimensionality of the current data to generate dimensionality-reduced data. The method includes generating a reconstruction error based on the dimensionality-reduced data and a plurality of neural network metrics. At least one of a size of the recent portion of the current data stored in the buffer or an amount of the reducing the dimensionality of the current data is based on the reconstruction error.

Abstract: Systems and methods are disclosed for anticipating a video switch to accommodate a new speaker in a video conference comprising a real time video stream captured by a camera local to a first videoconference endpoint is analyzed according to at least one speaker anticipation model. The speaker anticipation model predicts that a new speaker is about to speak. Video of the anticipated new speaker is sent to the conferencing server in response to a request for the video on the anticipated new speaker from the conferencing server. Video of the anticipated new speaker is distributed to at least a second videoconference endpoint.

Abstract: In one embodiment, a method includes receiving training data, the training data including training video data representing video of a location in a quiescent state, training a neural network using the training data to obtain a plurality of metrics, receiving current data, the current data including current video data representing video of the location at a current time period, generating a reconstruction error based on the plurality of metrics and the current video data in the embedded space, and generating, in response to determining that the reconstruction error is greater than a threshold, a notification indicative of the location being in a non-quiescent state.

Abstract: Various implementations disclosed herein provide a method for performing one or more transactions between application containers. In various implementations, the method includes transmitting a key request to a first network node within a cluster of network nodes that are configured to generate and maintain a distributed ledger. In some implementations, the key request indicates that the requested key is for one or more transactions between a first application container and a second application container. In various implementations, the method includes receiving a key in response to transmitting the key request. In some implementations, the key is valid for the one or more transactions between the first application container and the second application container. In various implementations, the method includes synthesizing, at the first application container, transaction data with the key.

Abstract: In one embodiment, a method includes obtaining text from a user, applying the text to a deep learning neural network to generate a plurality of bias coordinates defining a point in an embedded space, and, in response to determining that at least one of the plurality of bias coordinates exceeds a threshold, providing an indication of bias to the user.