Abstract: In some implementations, a device may receive a configuration associated with a machine learning model. The device may additionally receive a first hyperparameter set associated with the machine learning model. Accordingly, the device may estimate a first quantity of floating-point operations (FLOPs) associated with one or more epochs, for the machine learning model, based on the first hyperparameter set. The device may output, to a user, an indication of a first energy consumption associated with training the machine learning model based on the first quantity of FLOPs.

Abstract: Embodiments relate to a system, computer program product, and method for determining shelter areas for two-wheeler vehicles, and, more specifically, for dynamically distinguishing the behavior of two-wheeler vehicles and non-two-wheeler vehicles as an indicator of shelter areas from inclement weather. The behavior of the vehicles is distinguished through a plurality of two-wheeler vehicles slowing down and congregating at a particular location as a shelter against inclement weather, while non-two-wheeler vehicles may slow down, however, not stop proximate this location.

Abstract: Implementations include providing a robotic model representative of a robotic system, determining baseline data by executing a simulation using the robotic model and a simulation world that represents the environment, generating, by a perturbation system, perturbations to be included in a set of test cases, each test case including at least one perturbation to at least one object relative to a baseline of the at least one object determined from the baseline data, determining test result data for the set of test cases by modifying the at least one object in view of the at least one perturbation to provide a modified object and executing the simulation with the modified object, and displaying a graph representative of the at least one object based on the baseline data and test result data for a test case, the graph depicting variance of the object during the simulation using the test case.

Abstract: A method, system, and computer program product for determining selection parameters for filtering algorithms using environmental data of images. The method may include receiving an image. The method may also include analyzing the image using at least image processing. The method may also include identifying, based on the analyzing, image data and environmental data of the image. The method may also include inputting the image data and the environmental data into a machine learning algorithm, where the machine learning algorithm includes mapped relationships between at least the environmental data and selection parameters. The method may also include predicting, using the machine learning algorithm, optimal selection parameters for the image. The method may also include applying the optimal selection parameters to a filtering algorithm for the image. The system and computer program product may include similar steps.

Abstract: Embodiments relate to a system, computer program product, and method for determining shelter areas for two-wheeler vehicles, and, more specifically, for dynamically distinguishing the behavior of two-wheeler vehicles and non-two-wheeler vehicles as an indicator of shelter areas from inclement weather. The behavior of the vehicles is distinguished through a plurality of two-wheeler vehicles slowing down and congregating at a particular location as a shelter against inclement weather, while non-two-wheeler vehicles may slow down, however, not stop proximate this location.

Abstract: A computer-implemented method includes: determining, by a computer device, user behavior with an application; determining, by the computer device, a criticality of the application to a user; and changing, by the computer device, a hardcoded timeout parameter of the application into a dynamic timeout parameter based on the determined user behavior and the determined criticality of the application.

Abstract: An object detection algorithm is selectively applied to frames in a video. A frame in the video is analyzed using a set of neural networks of the object detection algorithm to detect a location of an object in the frame and predict a bounding box for the object in the frame. A magnitude of a delta between the frame and a second frame is determined. The magnitude of the delta is determined based on a difference between the two frames in values of at least one parameter of their respective sets of pixels. Responsive to the magnitude of the delta being less than a threshold, a new bounding box is predicted for the object in the second frame without analyzing the second frame using the set of neural networks.

Abstract: A method, system, and computer program product for determining selection parameters for filtering algorithms using environmental data of images. The method may include receiving an image. The method may also include analyzing the image using at least image processing. The method may also include identifying, based on the analyzing, image data and environmental data of the image. The method may also include inputting the image data and the environmental data into a machine learning algorithm, where the machine learning algorithm includes mapped relationships between at least the environmental data and selection parameters. The method may also include predicting, using the machine learning algorithm, optimal selection parameters for the image. The method may also include applying the optimal selection parameters to a filtering algorithm for the image. The system and computer program product may include similar steps.

Abstract: An example operation may include one or more of connecting, by a miner node, to a blockchain network comprised of a plurality user nodes, reading, by the miner node, a smart contract generated by a user node of the plurality of the user nodes, wherein the smart contract specifies a location of a user video file, acquiring, by the miner node, the user video file from the location, generating, by the miner node, a profile vector of the user video file based on properties of the user video file, determining, by the miner node, a cluster that has similar profile vectors, applying, by the miner node, a compression algorithm associated with the cluster to the user video file to generate a compressed user video file, and sending, by the miner node, the compressed user video file to the user node of the plurality of the user nodes for a validation.

Abstract: A method, and associated computer system and computer program product, for video compression that includes receiving a video file including a plurality of frames, identifying at least one image feature in each of the plurality of frames, determining a semantic state change of the image feature for each successive frame after a first of the plurality of frames, and storing the first of the plurality of frames and the semantic change of the image feature for each successive frame after the first of the plurality of frames.

Abstract: A processor(s) processes multiple image frames of a scene for object recognition by applying a first process of a plurality of visual recognition processes using different grid sizes, to one or more image frames. The first process divides the image frame(s) into an m×m grid to determine bounding boxes for identifying one or more objects. Based on detecting passage of a predefined time interval, less than a period of time represented by the image frames, processing dynamically switches to a second process to process one or more other image frames. The second process is applied to the other image frame(s), dividing the other image frame(s) into an n×n grid to determine bounding boxes for identifying one or more other objects in the other image frame(s), where the n×n grid used by the second process is different from the m×m grid.

Abstract: An object detection algorithm is selectively applied to frames in a video. A frame in the video is analyzed using a set of neural networks of the object detection algorithm to detect a location of an object in the frame and predict a bounding box for the object in the frame. A magnitude of a delta between the frame and a second frame is determined. The magnitude of the delta is determined based on a difference between the two frames in values of at least one parameter of their respective sets of pixels. Responsive to the magnitude of the delta being less than a threshold, a new bounding box is predicted for the object in the second frame without analyzing the second frame using the set of neural networks.

Abstract: A message analysis system controls the transmission of outgoing messages based on the contents of the outgoing messages so that the proliferation of restricted data or confidential content can be monitored or prevented. An outgoing message is detected via various user actions executed during the creation of the outgoing message. The contents of the outgoing message are extracted to determine if the outgoing message includes restricted data or content based on a comparison of the message contents with contents of an index which includes prior tagged messages and their message counts. If the contents of the outgoing message match a prior received message, an associated message count is updated and the outgoing message is transmitted or restricted based on the comparison of the updated message count with a corresponding count threshold.

Abstract: Approaches presented herein enable optimization of a developing application to a user base. More specifically, application-centric data is gathered during a cultivation phase of the developing application. Substantially concurrently with the cultivation phase of the developing application, the application-centric data is analyzed according to static code of the developing application, a testing of the developing application, or a user experience (UX) design of the developing application. A machine learning model is applied to the analyzed application-centric data. This machine learning model is trained on historic application feedback data from applications available to the user base. Based on the machine learning model, a recommended change to optimize the developing application to the user base is generated.

Abstract: Systems and methods for preventing software application tampering are disclosed. 1. In embodiments, a computer-implemented method includes: identifying, by a computing device, an IP related software code segment of a software application; segregating, by the computing device, the IP related software code segment into one or more native code clusters and one or more non-native code clusters; refactoring, by the computing device, the one or more non-native code clusters into one or more stand-alone portable components (SPCs); determining, by the computing device, code migration scores for the one or more SPCs; and determining, by the computing device, select SPCs from the one or more SPCs to migrate to a remote security server based on the code migration scores.

Abstract: A processor(s) processes multiple image frames of a scene for object recognition by applying a first process of a plurality of visual recognition processes using different grid sizes, to one or more image frames. The first process divides the image frame(s) into an m×m grid to determine bounding boxes for identifying one or more objects. Based on detecting passage of a predefined time interval, less than a period of time represented by the image frames, processing dynamically switches to a second process to process one or more other image frames. The second process is applied to the other image frame(s), dividing the other image frame(s) into an n×n grid to determine bounding boxes for identifying one or more other objects in the other image frame(s), where the n×n grid used by the second process is different from the m×m grid.

Abstract: Aspects of the present invention provide devices that filter a subset of items from a database according to a model which predicts sales of items in the database with social media data and the subset of items filtered according to a highest predicted sales. The model comprises a machine learning model trained with sales history of items from the database and social media data history. The devices load the filtered subset of items from the database into a data cache, receive a search which searches for one or more items in the database, and return one or more items of the loaded subset of items from the data cache in response to the received search.

Abstract: Approaches presented herein enable optimization of a developing application to a user base. More specifically, application-centric data is gathered during a cultivation phase of the developing application. Substantially concurrently with the cultivation phase of the developing application, the application-centric data is analyzed according to static code of the developing application, a testing of the developing application, or a user experience (UX) design of the developing application. A machine learning model is applied to the analyzed application-centric data. This machine learning model is trained on historic application feedback data from applications available to the user base. Based on the machine learning model, a recommended change to optimize the developing application to the user base is generated.

Abstract: A method and system for modifying a video file is provided. The method includes continuously receiving a video stream. A video frame of the video stream is divided into viewport portions each including an associated meta data portion. An eye focus of a user with respect to a first viewport portion is detected and a resulting importance score is determined for each viewport portion. Viewing trajectories for viewing each viewport portion are determined and a first group of viewport portions and a second group of viewport portions are determined. A payoff matrix associated with the viewing trajectories is generated and a best possible decision for the user with respect to viewing the viewport portions is generated. A third group of viewport portions is determined and presented with respect to the best possible decision.

Abstract: An example operation may include one or more of identifying a query from a requesting entity, where the query requests access to one or more blockchains, converting the query to an expression tree, creating one or more expression tree variations based on the expression tree, the one or more expression tree variations provide one or more different expressions than the expression tree and a same result as the expression tree, determining access conformity between one or more expression tree variations and the expression tree, selecting an expression tree variation with a greatest conformity rating, performing the query using the expression tree variation with the greatest conformity rating, and providing query results to a requesting entity.