Abstract: A beamforming vector for a communication link between a first device and a second device may be determined by searching, based on a signal from the second device, a hierarchical beamforming codebook including a plurality of beamforming vectors occupying a plurality of levels. The plurality of beamforming vectors may include a first beamforming vector and a second beamforming vector occupying a first level of the hierarchical beamforming codebook. The first beamforming vector may have a first angular range while the second beamforming vector may have a second angular range. The first beamforming vector may be selected as the beamforming vector based on a respective posterior probability an angle-of-arrival of the signal from the second device being in each of the first angular range and the second angular range. Related systems and articles of manufacture, including computer program products, are also provided.

Abstract: A method for localizing a target object may include applying, to a first image captured by an apparatus, a first machine learning model trained to determine whether the target object is visible in the first image depicting a first region of a search area. In response to the target object being absent from the first image, a second machine learning model may be applied to identify, based on the target object as being visible and/or nonvisible in one or more portions of the first image, a second region of the search area corresponding to a sub-region within the first region of the search area having a highest probability of including the target object. A command may be sent to the apparatus to navigate the apparatus to the second region of the search area in order to capture a second image of the second region of the search area.

Abstract: A method for detecting and/or preventing an adversarial attack against a target machine learning model may be provided. The method may include training, based at least on training data, a defender machine learning model to enable the defender machine learning model to identify malicious input samples. The trained defender machine learning model may be deployed at the target machine learning model. The trained defender machine learning model may be coupled with the target machine learning model to at least determine whether an input sample received at the target machine learning model is a malicious input sample and/or a legitimate input sample. Related systems and articles of manufacture, including computer program products, are also provided.

Abstract: Disclosed herein is a device for facilitating managing of paths for unmanned vehicles. Further, the device is configured to be communicatively coupled with the unmanned vehicles. Further, the device comprises a communication device configured for receiving heterogeneous information from sensors and transmitting commands to unmanned vehicles. Further, the device comprises a processing device communicatively coupled with the communication device. Further, the processing device is configured for integrating the heterogeneous information, generating integrated information based on the integrating, analyzing the integrated information based on paths assigned to the unmanned vehicles for performing a mission, determining an adjustment required to be made in the paths by the plurality of unmanned vehicles, and generating the commands for the unmanned vehicles. Further, the device comprises a storage device communicatively coupled with the processing device.

Abstract: A beamforming vector for a communication link between a first device and a second device may be determined by searching, based on a signal from the second device, a hierarchical beamforming codebook including a plurality of beamforming vectors occupying a plurality of levels. The plurality of beamforming vectors may include a first beamforming vector and a second beamforming vector occupying a first level of the hierarchical beamforming codebook. The first beamforming vector may have a first angular range while the second beamforming vector may have a second angular range. The first beamforming vector may be selected as the beamforming vector based on a respective posterior probability an angle-of-arrival of the signal from the second device being in each of the first angular range and the second angular range. Related systems and articles of manufacture, including computer program products, are also provided.

Abstract: A method may include training, based on a first training data available at a first node in a network, a first local machine learning model. A first local belief of a parameter set of a global machine learning model may be updated based on the training of the first local machine learning model. A second local belief of the parameter set of the global machine learning model may be received from a second node in the network. The second local belief may have been updated based on the second node training a second local machine learning model. The second local machine learning model may be trained based on a second training data available at the second node. The first local belief may be updated based on the second local belief of the second node. Related systems and articles of manufacture, including computer program products, are also provided.

Abstract: A method for localizing a target object may include applying, to a first image captured by an apparatus, a first machine learning model trained to determine whether the target object is visible in the first image depicting a first region of a search area. In response to the target object being absent from the first image, a second machine learning model may be applied to identify, based on the target object as being visible and/or nonvisible in one or more portions of the first image, a second region of the search area corresponding to a sub-region within the first region of the search area having a highest probability of including the target object. A command may be sent to the apparatus to navigate the apparatus to the second region of the search area in order to capture a second image of the second region of the search area.

Abstract: A method for detecting and/or preventing an adversarial attack against a target machine learning model may be provided. The method may include training, based at least on training data, a defender machine learning model to enable the defender machine learning model to identify malicious input samples. The trained defender machine learning model may be deployed at the target machine learning model. The trained defender machine learning model may be coupled with the target machine learning model to at least determine whether an input sample received at the target machine learning model is a malicious input sample and/or a legitimate input sample. Related systems and articles of manufacture, including computer program products, are also provided.

Abstract: An end-to-end method is provided for scheduling connections for networks such as all-optical data centers, which have a zero in-network buffer and a non-negligible reconfiguration delay in which the rate of schedule reconfiguration is limited to minimize the impact of reduced duty-cycles and to ensure bounded delay without overly restricting the rate of monitoring and decision processes. The method decouples the rate of scheduling from the rate of monitoring. A scheduling algorithm for switches with a reconfiguration delay is used which is based on the well-known MaxWeight scheduling policy. The scheduling policy requires no prior knowledge of traffic load.

Abstract: Systems and methods according to present principles provide an architecture for data center networks with many, e.g., possibly up to thousands, top of rack (ToR) switches, by employing an architecture that relies on a separation of the data and the control planes. While the data is switched between the ToR switches in an all-optical high rate network, network state and control information is continuously transmitted and received from a central unit (also termed a control unit or centralized unit) over an ultra-low-latency wireless/wired network.

Abstract: An end-to-end method is provided for scheduling connections for networks such as all-optical data centers, which have a zero in-network buffer and a non-negligible reconfiguration delay in which the rate of schedule reconfiguration is limited to minimize the impact of reduced duty-cycles and to ensure bounded delay without overly restricting the rate of monitoring and decision processes. The method decouples the rate of scheduling from the rate of monitoring. A scheduling algorithm for switches with a reconfiguration delay is used which is based on the well-known MaxWeight scheduling policy. The scheduling policy requires no prior knowledge of traffic load.

Abstract: A data routing mechanism is disclosed. The data routing mechanism includes virtual output queues (VOQs), corresponding to a first input port, that store data to be sent to one of a first and second output port. The data routing mechanism includes VOQs, corresponding to a second input port, that store data to be sent to one of the first and second output port. The data routing mechanism includes a switch fabric that includes a plurality of buffers at crosspoints between the first and second input ports and the first and second output ports. The data routing mechanism includes a first input scheduler that transmits a first data from the VOQs corresponding to the first input port to one of the plurality of buffers based on lengths of the VOQs corresponding to the first input port and a credit state of the plurality of buffers.

Abstract: A data routing mechanism is disclosed. The data routing mechanism includes virtual output queues (VOQs), corresponding to a first input port, that store data to be sent to one of a first and second output port. The data routing mechanism includes VOQs, corresponding to a second input port, that store data to be sent to one of the first and second output port. The data routing mechanism includes a switch fabric that includes a plurality of buffers at crosspoints between the first and second input ports and the first and second output ports. The data routing mechanism includes a first input scheduler that transmits a first data from the VOQs corresponding to the first input port to one of the plurality of buffers based on lengths of the VOQs corresponding to the first input port and a credit state of the plurality of buffers.