Abstract: Devices, systems, and methods for software-defined cognitive networking for wireless communications are provided. An example method of wireless communication includes performing, at a first node of a plurality of nodes, multiple network interference measurements to generate a first local interference model, receiving, from a second node of the plurality of nodes, a second local interference model, combining, at the first node, the first local interference model and the second local interference model to generate a joint interference model, generating, based on the joint interference model, a plurality of interference parameters that characterize a communication channel between the first node and the second node, selecting, based on the plurality of interference parameters, an operating waveform from a plurality of waveforms such that a performance metric for a data communication from the first node to the second node exceeds a threshold, and performing, using the operating waveform, the data communication.

Abstract: Systems, computer programs, devices, and methods that enable ML-based vision processing for low-power, embedded, and/or real-time applications. In one exemplary embodiment, smart glasses use classifiers that are based on machine-learned (ML) patch relationships. The ML patch features are determined during an offline training process. The ML patch features are grouped into weak classifiers, strong classifiers, and detectors to progressively improve prediction accuracy. An object detection architecture uses triggering logic, search management, and a classification neural network to enable event-based searching, interest-based searching, and/or dynamic search control. In some cases, pre-processing may also be used to minimize the neural network complexity (e.g., pre-processing for scaling, rotations, translations, etc.).

Abstract: Systems, computer programs, devices, and methods that enable ML-based vision processing for low-power, embedded, and/or real-time applications. In one exemplary embodiment, smart glasses use classifiers that are based on machine-learned (ML) patch relationships. The ML patch features are determined during an offline training process. The ML patch features are grouped into weak classifiers, strong classifiers, and detectors to progressively improve prediction accuracy. An object detection architecture uses triggering logic, search management, and a classification neural network to enable event-based searching, interest-based searching, and/or dynamic search control. In some cases, pre-processing may also be used to minimize the neural network complexity (e.g., pre-processing for scaling, rotations, translations, etc.).

Abstract: Systems, computer programs, devices, and methods that enable ML-based vision processing for low-power, embedded, and/or real-time applications. In one exemplary embodiment, smart glasses use classifiers that are based on machine-learned (ML) patch relationships. The ML patch features are determined during an offline training process. The ML patch features are grouped into weak classifiers, strong classifiers, and detectors to progressively improve prediction accuracy. An object detection architecture uses triggering logic, search management, and a classification neural network to enable event-based searching, interest-based searching, and/or dynamic search control. In some cases, pre-processing may also be used to minimize the neural network complexity (e.g., pre-processing for scaling, rotations, translations, etc.).

Abstract: Systems, computer programs, devices, and methods that enable ML-based vision processing for low-power, embedded, and/or real-time applications. In one exemplary embodiment, smart glasses use classifiers that are based on machine-learned (ML) patch relationships. The ML patch features are determined during an offline training process. The ML patch features are grouped into weak classifiers, strong classifiers, and detectors to progressively improve prediction accuracy. An object detection architecture uses triggering logic, search management, and a classification neural network to enable event-based searching, interest-based searching, and/or dynamic search control. In some cases, pre-processing may also be used to minimize the neural network complexity (e.g., pre-processing for scaling, rotations, translations, etc.).

Abstract: Devices, systems, and methods for software-defined cognitive networking for wireless communications are provided. An example method of wireless communication includes performing, at a first node of a plurality of nodes, multiple network interference measurements to generate a first local interference model, receiving, from a second node of the plurality of nodes, a second local interference model, combining, at the first node, the first local interference model and the second local interference model to generate a joint interference model, generating, based on the joint interference model, a plurality of interference parameters that characterize a communication channel between the first node and the second node, selecting, based on the plurality of interference parameters, an operating waveform from a plurality of waveforms such that a performance metric for a data communication from the first node to the second node exceeds a threshold, and performing, using the operating waveform, the data communication.