Techniques and systems are described for performing video segmentation using fully connected object proposals. For example, a number of object proposals for a video sequence are generated. A pruning step can be performed to retain high quality proposals that have sufficient discriminative power. A classifier can be used to provide a rough classification and subsampling of the data to reduce the size of the proposal space, while preserving a large pool of candidate proposals. A final labeling of the candidate proposals can then be determined, such as a foreground or background designation for each object proposal, by solving for a posteriori probability of a fully connected conditional random field, over which an energy function can be defined and minimized.

Hidden Features are locally extracted from Industrial Data of the industrial system by a Local Application executed on a local computer of a customer. The Hidden Features are uploaded to an external computer of a service provider. A Domain Model for the industrial system is externally determined from an Industrial Model Library (IML) on the external computer based on the uploaded Hidden Features by an External Algorithm including at least one Machine Learning Model (MLM) executed on the external computer. The determined Domain Model for the industrial system is provided to the customer. The at least one MLM has been trained on ranking most appropriate Domain Models for industrial systems based on Hidden Features of the respective industrial systems. The most appropriate Domain Models represent all relevant technical aspects of the respective industrial systems.

Systems and methods for analyzing machine learning of cognitive software robots (CogBots) over time are provided. In implementations, a method includes generating, by a computing device, a graph of historic learning curves based on historic learning data over time for a subject obtained from a primary cognitive software robot (CogBot) and at least one secondary CogBot; generating, by the computing device, a best probable learning curve based on the historic learning curves of the graph, wherein the best probable learning curve is predictive of future learning by the primary CogBot for the subject; and generating, by the computing device, information regarding a current status of the learning of the primary CogBot based on the best probable learning curve.

The playback of media by a playback device is controlled by input gestures. Each user gesture can be first broken down into a base gesture which indicates a specific playback mode. The gesture is then broken down into a second part which contains a modifier command which determines the speed for the playback mode determined from the base command. Media content is then played using the specified playback mode at a speed determined by the modifier command.

An estimation device (10) includes an input unit (30) and an estimation unit (32). A first observation value (40) for each of a plurality of observation areas (50), the first observation value being the number of presences (S) of persons who are observation targets at each of a plurality of observation times, and a second observation value (42) for each of a plurality of observation points (52) included in any one of the plurality of observation areas (50), the second observation value being the number of passages (C) of the persons at each of the plurality of observation times, are input to the input unit (30). The estimation unit (32) estimate at least one of the number of passages (C) of the person at an arbitrary estimation time (48) at any one of the plurality of observation points (52) and the number of presences (S) of the person at the arbitrary estimation time (48) in any one of the plurality of observation areas (50) based on a constraint condition (G) satisfied between the first observation value (40) and the second observation value (42), the first observation value (40), and the second observation value (42).

According to various embodiments, a machine-learning based system for simulating tropical cyclones (TCs) and assessing TC risk is disclosed. The system includes a hierarchical Poisson genesis module configured to develop a Poisson regression and TC genesis simulation on a plurality of clustering grids. The system further includes an analog-wind track module configured to determine movement of a TC by both analog predictors formed by historical track patterns and current in situ wind. The system additionally includes a Markov intensity module configured to determine intensity change of the TC by considering three hidden discrete states of storm intensity change and associating each state with a probability distribution of intensity change.

Markov random field parameters are identified to use for covariance modeling of correlation between gradient terms of a loss function of the classifier. A subset of images are sampled, from a dataset of images, according to a normal distribution to estimate the gradient terms. Black-box gradient estimation is used to infer values of the parameters of the Markov random field according to the sampling. Fourier basis vectors are generated from the inferred values. An original image is perturbed using the Fourier basis vectors to obtain loss function values. An estimate of a gradient is obtained from the loss function values. An image perturbation is created using the estimated gradient. The image perturbation is added to an original input to generate a candidate adversarial input that maximizes loss in identifying the image by the classifier. The neural network classifier is queried to determine a classifier prediction for the candidate adversarial input.

A system may include a data acquisition hub connected to databases and sensors associated with locomotives, systems, or components of a train and configured to acquire real-time and historical configuration, structural, and operational data in association with inputs derived from real time and historical contextual data relating to a plurality of trains. The system may include a virtual system modeling engine configured to receive results of a non-destructive evaluation of a train component, simulate in-train forces, determine a predicted time of failure for the train component based on an evaluation of stresses that have already been applied to the component and expected future stresses, and implement repair, replacement, or operational protocols for the train component before or at a repair facility that will be reached by the train ahead of a predetermined minimum threshold time period before the predicted time of failure.

Provided is a process, including: obtaining a first training dataset, training a first machine-learning model on the first training dataset, obtaining a set of candidate question sequences, forming virtual subject-entity records, forming a second training dataset, training a second machine-learning model, and storing the adjusted parameters of the second machine-learning model in memory.

A method utilizing a digital twin instance in relation to food to query current/future properties thereof. A digital twin instance representing the food is generated from a digital twin template. The digital twin instance has assigned thereto, for a first target variable describing a food property, a mathematical model with a model parameter and an environmental parameter. The digital twin instance has a probability distribution for the model parameter of the first target variable. In the course of the handling of the food until it reaches a shop and/or at the shop, a measurement of the parameter is made, the values thereof being stored and assigned to the twin instance. The mathematical model of the first target variable, the probability distribution of the model parameter and the values of the environmental parameter are used to ascertain a probability distribution with respect to the target variable.