A method for scheduling a set of jobs for a plurality of machines. Each job is defined by at least one feature which characterizes a processing time of the job. If any of the machines is free, a job from of the set of jobs is selected to be carrying out by said machine and scheduled for said machine. The job is selected as follows: a Graph Neural Network receives as input the set of jobs and a current state of at least the machine which is free, the Graph Neural Network outputs a reward for the set of jobs if launched on the machines, which states are inputted into the Graph Neuronal Network, and the job for the free machine is selected depending on the Graph Neural Network output.

For manufacturing process control, closed-loop control is provided (18) based on a constrained reinforcement learned network (32). The reinforcement is constrained (22) to account for the manufacturing application. The constraints may be for an amount of change, limits, or other factors reflecting capabilities of the controlled device and/or safety.

The present invention relates to a method for producing a dental restoration, comprising the steps of generating (S101) a three-dimensional dataset for describing the spatial shape of the dental restoration in a blank; adding (S102) the spatial shape of the dental restoration to a dataset of the blank; and integrating (S103) spatial data for holding pins for fixing the dental restoration into the three-dimensional dataset of the blank by a machine learning algorithm (103).

The current disclosure describes techniques for managing vertical alignment or overlay in semiconductor manufacturing using machine learning. Alignments of interconnection features in a fan-out WLP process are evaluated and managed through the disclosed techniques. Big data and neural networks system are used to correlate the overlay error source factors with overlay metrology categories. The overlay error source factors include tool related overlay source factors, wafer or die related overlay source factors and processing context related overlay error source factors.

A semiconductor fabrication scheduling method includes creating a load scheduling data schema including facility data of product lots to be dispatched to a plurality of workstations; generating a load schedule profile using a load-balancing model and based on the load scheduling data schema, wherein the load-balancing model includes one or more objective functions and there is at least one weight factor in an objective function; generating a current load schedule based on the load schedule profile; dispatching the product lots to the plurality of workstations using the current load schedule to complete fabrication of the product lots; obtaining a set of current key performance indicators (KPIs) of the completed fabrication of the product lots; and automatically adjusting the weight factors of the objective functions of the load-balancing model based on the current KPIs using a big-data architecture to generate a next load schedule for next cycle of fabrication.

The disclosure provides population and non-population-based classifiers to categorize patients and cancers. The population-based classifiers disclosed integrate signatures, i.e., global scores related to the expression of genes in particular gene panels. The non-population-based classifiers are generated using machine-learning techniques (e.g., regression, random forests, or ANN). Each type of classifier stratifies patients and cancers according to tumor microenvironments (TME) as biomarker-positive or biomarker-negative, and treatment decisions are then guided by the presence/absence of a particular TME. Also provided are methods for treating a subject, e.g., a human subject, afflicted with cancer comprising administering a particular therapy depending on the classification of the cancer's TME according to the disclosed classifiers. Also provided are personalized treatments that can be administered to a subject having a cancer classified into a particular TME, and gene panels that can be used for identifying a human subject afflicted with a cancer suitable for treatment with a particular therapeutic agent.

In some embodiments, identifying a replay attack in an industrial control system of an industrial asset includes receiving a first set of time series data associated with an ambient condition of one or more first monitoring nodes at a first location of the industrial control system. An actual system feature value for the industrial asset is determined based upon the first set of time series data. A second set of time series data indicative of the ambient condition at a second location is received, and a nominal system feature value is determined based upon the second set of time series data. A correlation between the actual feature value and the nominal system feature value is analyzed to determine a correlation result. A request received by the industrial control system is selectively categorized as a replay attack based upon the correlation result.

Systems and methods are described herein for real-time data processing and for emergency planning. Scenario test data may be collected in real-time based on monitoring local or regional data to ascertain any anomaly phenomenon that may indicate an imminent danger or of concern. A computer-implemented method may include filtering a plurality of different test scenarios to identify a sub-set of test scenarios from the plurality of different test scenarios that may have similar behavior characteristics. A sub-set of test scenarios is provided to a trained neural network to identify one or more sub-set of test scenarios. The one or more identified sub-set of test scenarios may correspond to one or more anomaly test scenarios from the sub-set of test scenarios that is most likely to lead to an undesirable outcome. The neural network may be one of: a conventional neural network and a modular neural network.

A method and system of controlling an appliance includes: receiving, from a first home appliance, a request to start video image processing for detecting a motion gesture of a user; processing a sequence of image frames captured by a camera corresponding to the first home appliance to identify a first motion gesture; selecting a second home appliance as a target home appliance for the first motion gesture in accordance with one or more target selection criteria, including first target selection criteria based on a location of the user relative to the first home appliance and second target selection criteria based on a level of match between the first motion gesture and a first control gesture corresponding to the second home appliance; and generating a control command to control the second home appliance in accordance with the first control gesture corresponding to the second home appliance.

An image-based Artificial Neural Networks (ANN) is used for photomask modeling, which can self-construct an internal representation of the photomask manufacturing process, therefore allowing the modeling process to become unfettered by the limitations of existing mathematical/statistical tools, thus greatly reduces/eliminates the effort needed from tedious and costly model-builders. The ANN model requires mask layout data converted into image pixel form. In ANN training phase a first circuit image and its existing SEM image are modeled via multiple layers of convolution and rectification to pick out the salient features of transformed image. In ANN testing phase, a second circuit image and its existing SEM image are compared and verified to have a difference smaller than the predetermined requirement. The satisfactory second circuit image is converted back from pixel form to circuit layout data for photomask writing.