Sri International

Amazon Technologies, Inc.

Dejan Jovanovic

Neha Rungta

Gabriela F Ciocarlie

Mohamed R Amer

Amit Goel

Michael E Locasto

Timothy J Meo

Alex C Tozzo

Yang Hu

Cherita Corbett

Disclosed are systems and methods for adjusting the encoding/ordering of parts of a satisfiability modulo theories ('SMT') problem indicated in an SMT configuration and/or optimizing the SMT solver configuration of an SMT solver to reduce the processing time needed to process the SMT problem using the SMT solver. For example, the order of parts of an SMT problem may be rearranged to reduce a processing time of the SMT problem. Likewise, the configuration options selected for an SMT solver may be adjusted to reduce the processing time required to process an SMT problem.

Adjusting satisfiability modulo theories solver configurations

Intrusion detection systems and methods monitor legal control messages in an operational control system to detect subtly malicious sequences of control messages with undesirable emergent effects on devices in the operational control system. A message provenance component may investigate system-level correlations between messages rather than detecting if individual messages are anomalous. A semantic fuzzing component may search, based on the operational effect of candidate message sequences, the space of legal messages for sequences that cause actual harm. Behavior oracles may be used to test message sequences to identify sequences that induce drift towards a failure state. The intrusion detection system is able to prevent harm and disruption arising from control messages that individually appear legitimate and benign but that, in combination with other messages, can cause undesirable outcomes.

Intrusion detection via semantic fuzzing and message provenance

This disclosure describes techniques that include generating, based on a description of a scene, a movie or animation that represents at least one possible version of a story corresponding to the description of the scene. This disclosure also describes techniques for training a machine learning model to generate predefined data structures from textual information, visual information, and/or other information about a story, an event, a scene, or a sequence of events or scenes within a story. This disclosure also describes techniques for using GANs to generate, from input, an animation of motion (e.g., an animation or a video clip). This disclosure also describes techniques for implementing an explainable artificial intelligence system that may provide end users with information (e.g., through a user interface) that enables an understanding of at least some of the decisions made by the AI system.

Artificial intelligence for generating structured descriptions of scenes

Growing community of inventors

Brooklyn, NY, United States of America

Dejan Jovanovic