Abstract: A system may configure a pipeline for a target device. The pipeline may include a signal processing component and a machine learning component. The pipeline may be configured to receive input data and generate output data based on the input data. For example, the output data may indicate detections in an output stream based on events in the input data in an input stream. The system may determine multiple post-processing configurations for post-processing the output data. A post-processing configuration may be configured to generate a detectable event based on the output data. The multiple post-processing configurations may be generated using a multi-objective optimization that varies one or more parameters for generating the detectable event.

Abstract: An object detection system may be configured for an embedded device, such as a microcontroller. The object detection system may be configured to divide an image into multiple cells arranged in a grid. A cell of the multiple cells may map to a region of one or more pixels in the image. The object detection system may detect in each cell either a background or one of multiple objects that are detectable classes distinct from one another. The background may be detected when none of the multiple objects are detected. The one of the multiple objects may be detected when a centroid of the one of the multiple objects is detected. In some implementations, expected data may be detected in a cell when detecting the data within a range, and anomalous data may be detected in a cell when detecting the data outside of the range.

Abstract: An anomaly detection system may be configured for an embedded device, such as a microcontroller. The anomaly detection system may be configured to receive a dataset from a sensor, such as a camera, a microphone, or an inertial management unit. The anomaly detection system may extract a plurality of features from the dataset. The plurality of features may be configured to train a neural network model, such as a convolutional classifier, to generate one or more classifications. The anomaly detection system may generate an anomaly score based on the plurality of features. The anomaly detection system may trigger an output based on the anomaly score exceeding a range.

Abstract: An input indicating a target device may be received. A processor may execute instructions stored in memory to determine performances of multiple configurations of a pipeline. The pipeline may include a signal processing component and a machine learning component. A configuration of the multiple configurations may vary one or more parameters for configuring the signal processing component or the machine learning component. A performance of a configuration of the multiple configurations may be determined based on the target device, indicated by the input, for implementing the configuration. In some implementations, determining a performance of a configuration may include calculating a latency, a memory usage, an energy usage, or an accuracy associated with the configuration when implemented on the target device.