Abstract: A system for inferring features in a digital image includes a vision sensor for capturing and optically encrypting an image of a visual scene so as to shroud any human-perceivable features contained therein. An image interpreter in communication with the vision sensor is programmed to recognize and classify features in the encrypted image without prior decryption, thereby ensuring the privacy of any subjects depicted therein. The vision sensor is encoded with a calibration key that applies a unique transformation function to optically encrypt the captured image. The image interpreter is permitted access to the calibration key and applies a convolutional network constructed using the calibration key, a source convolutional network, and a set of private layer keys in order to infer features in the encrypted image without prior decryption. For applications requiring enhanced privacy protection, the inference output produced by the image interpreter remains encrypted but available for further feature detection.

Abstract: A communication network authenticates the source of messages transmitted on a flat bus to determine the presence of spoofing events. A programmable intrusion detection device is connected to the bus at a fixed location and compiles templates for various tri-bit signal pulses that form the data transmitted as messages between network nodes. Each tri-bit template compares unique signal characteristics inherent in the signal waveform received by the device from each node, the unique characteristics being directly attributable to the physical topology of the network. In use, the device uses the templates to calculate an inferred source identifier for each message. The inferred source identifier is then compared against the declared source identifier, which is embedded in message metadata, to authenticate the message source. Any lack of reconciliation between the inferred and declared source identifiers causes the device to mark the message as spoofed and initiate a designated response.

Abstract: A system for discovering internet memes from a social media stream includes a meme detector for grouping together incoming posts that contain a common set of anomalous terms, the meme detector comparing the frequency of use of the anomalous terms against normalcy models to ensure that the use of the anomalous terms is non-recurring. The system additionally includes a meme tracker for classifying each cluster of related posts as a particular meme and, in turn, associating new related posts from the media stream with the classified cluster. In this manner, the particular set of anomalous terms used to classify each meme can be augmented to include other discriminating terms. Accordingly, the parameters, or features, associated with each meme are continuously monitored and refined, as needed, to ensure that the most comprehensive and reflective cluster of posts for each meme is collected from the social media stream in an efficient fashion.

Abstract: A system for discovering internet memes from a social media stream includes a meme detector for grouping together incoming posts that contain a common set of anomalous terms, the meme detector comparing the frequency of use of the anomalous terms against normalcy models to ensure that the use of the anomalous terms is non-recurring. The system additionally includes a meme tracker for classifying each cluster of related posts as a particular meme and, in turn, associating new related posts from the media stream with the classified cluster. In this manner, the particular set of anomalous terms used to classify each meme can be augmented to include other discriminating terms. Accordingly, the parameters, or features, associated with each meme are continuously monitored and refined, as needed, to ensure that the most comprehensive and reflective cluster of posts for each meme is collected from the social media stream in an efficient fashion.

Abstract: A system for reconstructing broadband analog signals from digital information includes a digital processing unit for decomposing a digital input signal into a plurality of multirate, digital input signal sub-components with optimized quantization levels, and a digital-to-analog converter (DAC) bank for converting the plurality of digital input signal sub-components into a corresponding set of analog signal sub-components, which are then combined and time-aligned. The digital processing unit preferably includes a multirate filterbank for separating the digital input signal through discrete wavelet decomposition, such through Haar wavelet decomposition. The DAC bank includes an array of digital-to-analog converters (DACs) with sample rates that vary by a constant factor.