Abstract: A method performed by one or more computers for obtaining an optimized algorithm that (i) is functionally equivalent to a target algorithm and (ii) optimizes one or more target properties when executed on a target set of one or more hardware devices. The method includes: initializing a target tensor representing the target algorithm; generating, using a neural network having a plurality of network parameters, a tensor decomposition of the target tensor that parametrizes a candidate algorithm; generating target property values for each of the target properties when executing the candidate algorithm on the target set of hardware devices; determining a benchmarking score for the tensor decomposition based on the target property values of the candidate algorithm; generating a training example from the tensor decomposition and the benchmarking score; and storing, in a training data store, the training example for use in updating the network parameters of the neural network.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for processing inputs using a neural network system that includes one or more transformed activation function layers.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for executing depth-parallel training of a neural network. One of the methods includes receiving an input sequence; and at each processing time step in a sequence of processing time steps: processing an input item using a first layer block in a stack of layer blocks to generate a first block output; for each subsequent layer block, processing a block output generated by the preceding layer block at the preceding processing time step to generate a current block output; computing i) a current error in an output item generated by the final layer block and ii) a current gradient of the current error; generating a parameter update for the final layer block; for each particular layer block that is not the final layer block, computing a current gradient for the particular layer block and generating a parameter update.

Abstract: A vehicle weight detection method, system, and non-transitory computer readable medium, include calculating a first difference between a first expected weight of a vehicle and a first current weight of the vehicle based on On-Board Dash (OBD) input data, calculating a second difference between a second expected weight of the vehicle and a second current weight of the vehicle based on a spring-mass-damper mechanical algorithm using vehicle data received from a user device, and a comparing circuit configured to compare each of the first difference and the second difference to a predetermined weight difference threshold value.

Abstract: A geolocation data discovery method, system, and non-transitory computer readable medium, include mining a textual alert message of an event from secure source data for a geolocational indicator, a temporal indicator, and a type of alert, embedding a geolocational tag location and a temporal tag time to the event based on the geolocational indicator and the temporal indicator mined from the textual alert message by the secure text mining, mapping the event with an icon related to the type of alert at the geolocational tag location on a navigational map at a time associated with the temporal tag time of the event, learning a false-positive event by comparing a secure sensor with an unsecure sensor as the secure source data, and removing updates to the icon on the map based on the learned false-positive event.

Abstract: A geolocation data discovery method, system, and non-transitory computer readable medium, include mining a textual alert message of an event from secure source data for a geolocational indicator, a temporal indicator, and a type of alert, embedding a geolocational tag location and a temporal tag time to the event based on the geolocational indicator and the temporal indicator mined from the textual alert message by the secure text mining, mapping the event with an icon related to the type of alert at the geolocational tag location on a navigational map at a time associated with the temporal tag time of the event, learning a false-positive event by comparing a secure sensor with an unsecure sensor as the secure source data, and removing updates to the icon on the map based on the learned false-positive event.

Abstract: A geolocation data discovery method, system, and non-transitory computer readable medium, include a secure text mining circuit configured to mine a textual alert message of an event from secure source data for a geolocational indicator, a temporal indicator, and a type of alert, a geolocation and temporal embedding circuit configured to embed a geolocational tag location and a temporal tag time to the event based on the geolocational indicator and the temporal indicator mined from the textual alert message by the secure text mining circuit, and an event mapping circuit configured to map the event with an icon related to the type of alert at the geolocational tag location on a navigational map at a time associated with the temporal tag time of the event.

Abstract: A vehicle weight detection method, system, and non-transitory computer readable medium, include calculating a first difference between a first expected weight of a vehicle and a first current weight of the vehicle based on On-Board Dash (OBD) input data, calculating a second difference between a second expected weight of the vehicle and a second current weight of the vehicle based on a spring-mass-damper mechanical algorithm using vehicle data received from a user device, a comparing circuit configured to compare each of the first difference and the second difference to a predetermined weight difference threshold value, and a location checking circuit configured to checks if a location of the vehicle matches a location in a list of verified weight change locations of a database if the comparing circuit detects at least one of the first difference and the second difference is greater than the predetermined weight difference threshold value.

Abstract: A geolocation data discovery method, system, and non-transitory computer readable medium, include a secure text mining circuit configured to mine a textual alert message of an event from secure source data for a geolocational indicator, a temporal indicator, and a type of alert, a geolocation and temporal embedding circuit configured to embed a geolocational tag location and a temporal tag time to the event based on the geolocational indicator and the temporal indicator mined from the textual alert message by the secure text mining circuit, and an event mapping circuit configured to map the event with an icon related to the type of alert at the geolocational tag location on a navigational map at a time associated with the temporal tag time of the event.

Abstract: A vehicle weight detection method, system, and non-transitory computer readable medium, include calculating a first difference between a first expected weight of a vehicle and a first current weight of the vehicle based on On-Board Dash (OBD) input data, calculating a second difference between a second expected weight of the vehicle and a second current weight of the vehicle based on a spring-mass-damper mechanical algorithm using vehicle data received from a user device, a comparing circuit configured to compare each of the first difference and the second difference to a predetermined weight difference threshold value, and a location checking circuit configured to checks if a location of the vehicle matches a location in a list of verified weight change locations of a database if the comparing circuit detects at least one of the first difference and the second difference is greater than the predetermined weight difference threshold value.

Abstract: A system and method for analyzing a memory element includes modeling the memory element using a simulation method and determining component response characteristics for components of the memory element. Safety regions are computed in a state space of the memory element, which indicate stable states. A transient analysis is performed to determine a path and time needed to reach one of the safety regions. Based on the path and time needed to reach one of the safety regions, a clock waveform or waveforms are determined which place a corresponding state in that safety region.

Abstract: A system and method for analyzing a memory element includes modeling the memory element using a simulation method and determining component response characteristics for components of the memory element. Safety regions are computed in a state space of the memory element, which indicate stable states. A transient analysis is performed to determine a path and time needed to reach one of the safety regions. Based on the path and time needed to reach one of the safety regions, a clock waveform or waveforms are determined which place a corresponding state in that safety region.

Abstract: A method of scheduling delivery of at least one of goods (e.g., repair parts) and services (e.g., repair services) includes inputting a performance objective (e.g., a plurality of performance objectives), and network data pertaining to a network of mobile points, and generating (e.g., by utilizing the input network data and the performance objective) a delivery schedule for a delivery vehicle (e.g., one or more delivery vehicles) to deliver the at least one of the goods and services.