Abstract: There is provided a computer implemented method of training a detection machine learning model (ML) for identifying a specific deepfake tool of a plurality of deepfake tools used to create a deepfake video, comprising: feeding a plurality of sample authentic videos into the specific deepfake tool, obtaining a plurality of deepfake videos as an outcome of the specific deepfake tool, creating a training dataset comprising a plurality of records, wherein a record includes a deepfake video labelled with a ground truth indicating the specific deepfake tool used to create the deepfake video, and training the detection ML model on the training dataset for detecting that the specific deepfake tool was used to create an input deepfake video.

Abstract: There is provided a computer implemented method of detecting a deepfake video, comprising: analyzing a video, annotating the video according to the analysis, selecting a plurality of verification models according to the annotations, wherein the plurality of verification models include at least one forensic model that is not trained to detect that the video comprises a deepfake video, wherein outcome of the at least one forensic model applied to the video is compared to a ground truth obtained by applying the at least one forensic model to at least one other authentic video having at least one annotation of the video, aggregating a plurality of outcomes of the plurality of verification models, and determining a probability that the video was created by a deepfake tool according to the aggregation.

Abstract: There is provided a computer implemented method of training a customized detector ML model for use on an input video, comprising: analyzing the input video to identify a plurality of features associated with the input video, obtaining a plurality of authentic videos associated with the plurality of features, feeding the plurality of authentic videos into at least one deepfake tool, obtaining a plurality of deepfake videos as an outcome of the at least one deepfake tool, creating a training dataset comprising a plurality of records, wherein at least one first record includes a deepfake video labelled with a ground truth indicating deepfake, and at least one second record includes an authentic video labelled with a ground truth indicating authentic, and training the customized detector ML model on the training dataset.

Abstract: A computer implemented method of tracking a travelling vessel, comprising obtaining a list of plurality of satellites capable of detecting the vessel at location(s) along predicted path(s) of the vessel. For each of the location(s) the following is performed: (a) Predicting vessel's possible future location(s) according to estimated movement vectors derived from a movement graph generated based on historical movement path(s), a recent movement path and a current location of the vessel. (b) Estimating satellites observation windows to identify candidate observation window(s) in which the satellite(s) have visual coverage of the possible future location(s). (c) Calculating detection score for each candidate observation window according to location probability score assigned to the possible future locations and view probability score assigned to the candidate observation windows. (d) Selecting preferred observation window presenting highest detection score.

Abstract: A computer implemented method of tracking a travelling vessel, comprising obtaining a list of plurality of satellites capable of detecting the vessel at location(s) along predicted path(s) of the vessel. For each of the location(s) the following is performed: (a) Predicting vessel's possible future location(s) according to estimated movement vectors derived from a movement graph generated based on historical movement path(s), a recent movement path and a current location of the vessel. (b) Estimating satellites observation windows to identify candidate observation window(s) in which the satellite(s) have visual coverage of the possible future location(s). (c) Calculating detection score for each candidate observation window according to location probability score assigned to the possible future locations and view probability score assigned to the candidate observation windows. (d) Selecting preferred observation window presenting highest detection score.

Abstract: A device, system, and method for solving a plurality of sub-problems representing a main problem pertaining to an environment. An abstract graph corresponding to each of the sub-problems may be constructed in a computing system for each of the sub-problems. At least a first abstract graph of the abstract graphs may be resolved, for example, by decomposing one or more nodes of the first abstract graph. At least one of the one or more nodes of the first abstract graph may be associated with one or more nodes of others of the abstract graphs. At least one of the abstract graphs may be resolved using information obtained by the association. Other embodiments are described and claimed.

Abstract: A device, system, and method for solving a plurality of sub-problems representing a main problem pertaining to an environment. An abstract graph corresponding to each of the sub-problems may be constructed in a computing system for each of the sub-problems. At least a first abstract graph of the abstract graphs may be resolved, for example, by decomposing one or more nodes of the first abstract graph. At least one of the one or more nodes of the first abstract graph may be associated with one or more nodes of others of the abstract graphs. At least one of the abstract graphs may be resolved using information obtained by the association. Other embodiments are described and claimed.