Abstract: Systems and methods are provided for determining an order of execution of a neural network. For instance, data indicative of a neural network and data indicative of an amount of available memory in a constrained memory space can be obtained. The neural network can include a plurality of operators. An order of execution associated with the neural network can then be determined. The order of execution specifies an order in which to execute each of the plurality of operators. The order of execution is determined based at least in part on the available memory in the constrained memory space. In particular, one or more graph search algorithms can be performed on a graph that is representative of the neural network to obtain the order of execution.

Abstract: Provided are systems and methods for a tour optimization application programming interface (API). The API can include a first set of instructions associated with a tour optimization data structure that can be associated with the API and specify a plurality of messages including one or more fields. The plurality of messages can be associated with a set of shipments and a set of vehicles. A second set of instructions can be associated with a modeling function that can implement one or more calls to generate model data based in part on the plurality of messages. A third set of instructions can be associated with implementation of a tour optimization service and can generate routing data for a user of the software application based on the model data. The routing data can include a set of routes based in part on the set of shipments and the set of vehicles.

Abstract: Systems and methods are provided for determining an order of execution of a neural network. For instance, data indicative of a neural network and data indicative of an amount of available memory in a constrained memory space can be obtained. The neural network can include a plurality of operators. An order of execution associated with the neural network can then be determined. The order of execution specifies an order in which to execute each of the plurality of operators. The order of execution is determined based at least in part on the available memory in the constrained memory space. In particular, one or more graph search algorithms can be performed on a graph that is representative of the neural network to obtain the order of execution.

Abstract: A computer-implemented method of constructing a transit transfer network from transit data comprises processing transit data to define a plurality of transfer endpoints; determining a set of acceptable transfer durations for each of a plurality of transfers associated with the transfer endpoints; determining one or more groups of endpoints to be merged, each group comprising two or more endpoints, wherein transfers associated with the endpoints in each said group have one or more common acceptable transfer durations; and merging the endpoints in each group to form one or more merged endpoints.

Abstract: Computer-implemented methods for scoring and ranking transit stops and transit lines are provided. In one aspect, a method includes determining data and factors associated with a first transit stop or line and calculating a score for the first transit stop or line. The data and factors used in the scoring calculation may be a combination of the number of vehicle departures from the transit stop, the number of transit lines connected to the transit stop, a point of interest proximity factor, a population center factor, the number of trips on the transit line, the number of transit stops on the transit line, the number of interconnected transit lines and the travel distance of the transit line. The transit stop and transit lines scores may be iterated over each other. Systems and machine-readable media for scoring and ranking transit stops and transit lines are also provided.

Abstract: A public transit travel planning system and methodology that uses an extensive pre-processing approach of transit information prior to query time on order to determine optimal public transit routes for journeys. At query time, since the transit information has already been processed by the system, very little computation is needed in order to fulfill the query. The system then provides users with public transit directions in response to the queries for public transit journeys.

Abstract: A public transit travel planning system and methodology that uses an extensive pre-processing approach of transit information prior to query time on order to determine optimal public transit routes for journeys. At query time, since the transit information has already been processed by the system, very little computation is needed in order to fulfill the query. The system then provides users with public transit directions in response to the queries for public transit journeys.

Abstract: A public transit travel planning system and methodology that uses an extensive pre-processing approach of transit information prior to query time on order to determine optimal public transit routes for journeys. At query time, since the transit information has already been processed by the system, very little computation is needed in order to fulfill the query. The system then provides users with public transit directions in response to the queries for public transit journeys.

Abstract: A public transit travel planning system and methodology that uses an extensive pre-processing approach of transit information prior to query time on order to determine optimal public transit routes for journeys. At query time, since the transit information has already been processed by the system, very little computation is needed in order to fulfill the query. The system then provides users with public transit directions in response to the queries for public transit journeys.

Abstract: A public transit travel planning system and methodology that uses an extensive pre-processing approach of transit information prior to query time on order to determine optimal public transit routes for journeys. At query time, since the transit information has already been processed by the system, very little computation is needed in order to fulfill the query. The system then provides users with public transit directions in response to the queries for public transit journeys.