Abstract: During a round of decentralized learning for updating of a global machine learning (ML) model, remote processor(s) of a remote system may transmit, to a population of computing devices, primary weights for a primary version of the global ML model, and cause each of the computing devices to generate a corresponding update for the primary version of the global ML model. Further, the remote processor(s) may cause the primary version of the global ML model to be updated based on the corresponding updates that are received during the round of decentralized learning. However, the remote processor(s) may receive other corresponding updates subsequent to the round of decentralized learning. Accordingly, various techniques described herein (e.g., FARe-DUST, FeAST on MSG, and/or other techniques) enable the other corresponding updates to be utilized in achieving a final version of the global ML model.

Abstract: Implementations disclosed herein are directed to various techniques for mitigating and/or preventing catastrophic forgetting in federated learning of global machine learning (ML) models. Implementations may identify a global ML model that is initially trained at a remote server based on a server data set, determine server-based data for global weight(s) of the global ML model, and transmit the global ML model and the server-based data to a plurality of client devices. The server-based data may include, for example, EWC loss term(s), client augmenting gradients, server augmenting gradients, and/or server-based data. Further, the plurality client devices may generate, based on processing corresponding predicted output and using the global ML model, and based on the server-based data, a corresponding client gradient, and transmit the corresponding client gradient to the remote server. Implementations may further generate an updated global ML model based on at least the corresponding client gradients.

Abstract: Implementations disclosed herein are directed to utilizing elastic weight consolidation (EWC) loss term(s) in federated learning of global machine learning (ML) models. Implementations may identify a global ML model that initially trained at a remote server based on a server data set, determine the EWC loss term(s) for global weight(s) of the global ML model, and transmit the global ML model and the EWC loss term(s) to a plurality of client devices. The EWC loss term(s) may be determined based on a Fisher information matrix for the server data set. Further, the plurality client devices may generate, based on processing corresponding predicted output and using the global ML model, and based on the EWC loss term(s), a corresponding client gradient, and transmit the corresponding client gradient to the remote server. Implementations may further generate an updated global ML model based on at least the corresponding client gradients.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. The objects can comprise satellites in a constellation of satellites. In some implementations, the satellites are imaging satellites, and the systems and methods for scheduling can use human collaboration to determine events of interest for acquisition of images. In some implementations, dominant edges are removed from the directed acyclic graph. In some implementations, dynamic weights are assigned to nodes associated with downlink events in the directed acyclic graph.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. The objects can comprise satellites in a constellation of satellites. In some implementations, the satellites are imaging satellites, and the systems and methods for scheduling can use human collaboration to determine events of interest for acquisition of images. In some implementations, dominant edges are removed from the directed acyclic graph. In some implementations, dynamic weights are assigned to nodes associated with downlink events in the directed acyclic graph.

Abstract: Systems and methods are provided for calculating satellite access windows for a constellation of imaging satellites. In some implementations, systems and methods are provided for managing execution of native programs on high performance computing systems.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. The objects can comprise satellites in a constellation of satellites. In some implementations, the satellites are imaging satellites, and the systems and methods for scheduling can use human collaboration to determine events of interest for acquisition of images. In some implementations, dominant edges are removed from the directed acyclic graph. In some implementations, dynamic weights are assigned to nodes associated with downlink events in the directed acyclic graph.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. The objects can comprise satellites in a constellation of satellites. In some implementations, the satellites are imaging satellites, and the systems and methods for scheduling can use human collaboration to determine events of interest for acquisition of images. In some implementations, dominant edges are removed from the directed acyclic graph. In some implementations, dynamic weights are assigned to nodes associated with downlink events in the directed acyclic graph.

Abstract: Systems and methods are provided for selecting, pricing, and prioritizing images obtained by a constellation of imaging satellites. The systems and methods presented can determine whether an imagery collection request should be accepted or rejected. If the imagery collection request is accepted, the systems and methods presented can determine an appropriate pricing option for the imagery collection request and how the imagery collection request should be prioritized in relation to other outstanding imagery collection requests.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. Violation of cumulative constraints can be identified at the moment of constraint violation such that events resulting in constraint violations can be removed from the schedule while the schedule is being determined. By removing the events triggering constraint violations at the point of constraint violation, the systems and methods provided can determine optimal or near-optimal schedules in a relatively quick and efficient manner compared to systems and methods that check for violations of cumulative constraints after determining a schedule. The objects can comprise satellites in a constellation of satellites.

Abstract: Systems and methods are provided for calculating satellite access windows for a constellation of imaging satellites. In some implementations, systems and methods are provided for managing execution of native programs on high performance computing systems.

Abstract: Systems and methods are provided for calculating satellite access windows for a constellation of imaging satellites. In some implementations, systems and methods are provided for managing execution of native programs on high performance computing systems. In one embodiment a system can determine, for each time interval within a first period of time, a position of each imaging satellite of a constellation of imaging satellites. The system can transform, for each time interval within the first period of time, the position of each imaging satellite from a first coordinate system to a second coordinate system. The system can determine an access window for at least one imagining satellite based at least in part on a determined angle between the vector to the respective location and a determined vector to the respective satellites. The system can schedule the at least one imaging satellite to perform a task within the access window.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. Violation of cumulative constraints can be identified at the moment of constraint violation such that events resulting in constraint violations can be removed from the schedule while the schedule is being determined. By removing the events triggering constraint violations at the point of constraint violation, the systems and methods provided can determine optimal or near-optimal schedules in a relatively quick and efficient manner compared to systems and methods that check for violations of cumulative constraints after determining a schedule. The objects can comprise satellites in a constellation of satellites.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. Violation of cumulative constraints can be identified at the moment of constraint violation such that events resulting in constraint violations can be removed from the schedule while the schedule is being determined. By removing the events triggering constraint violations at the point of constraint violation, the systems and methods provided can determine optimal or near-optimal schedules in a relatively quick and efficient manner compared to systems and methods that check for violations of cumulative constraints after determining a schedule. The objects can comprise satellites in a constellation of satellites.

Abstract: Examples of satellite scheduling systems are provided that use crowd-sourced data to generate image acquisition events for a network of imaging satellites. A crowd-sourcing system may utilize crowd-sourced data (e.g., messages generated by users of social network services) to determine events of interest and geographic locations of such events. Event data may then be used to create or update image acquisition tasks and/or task priorities which are automatically provided to a scheduling system to facilitate timely and responsive acquisition of overhead images of the geographic location of the event. The scheduling system can utilize a directed acyclic graph to increase or maximize a utilization function, which can lead to determination of optimal or near-optimal schedules in a relatively quick and efficient manner.

Abstract: Examples of satellite scheduling systems are provided that use crowd-sourced data to generate image acquisition events for a network of imaging satellites. A crowd-sourcing system may utilize crowd-sourced data (e.g., messages generated by users of social network services) to determine events of interest and geographic locations of such events. Event data may then be used to create or update image acquisition tasks and/or task priorities which are automatically provided to a scheduling system to facilitate timely and responsive acquisition of overhead images of the geographic location of the event. The scheduling system can utilize a directed acyclic graph to increase or maximize a utilization function, which can lead to determination of optimal or near-optimal schedules in a relatively quick and efficient manner.

Abstract: Systems and methods are provided for scheduling objects having pair-wise and cumulative constraints. The systems and methods presented can utilize a directed acyclic graph to increase or maximize a utilization function. Violation of cumulative constraints can be identified at the moment of constraint violation such that events resulting in constraint violations can be removed from the schedule while the schedule is being determined. By removing the events triggering constraint violations at the point of constraint violation, the systems and methods provided can determine optimal or near-optimal schedules in a relatively quick and efficient manner compared to systems and methods that check for violations of cumulative constraints after determining a schedule. The objects can comprise satellites in a constellation of satellites.