Abstract: Provided are computing systems, methods, and platforms that automatically produce production-ready machine learning models and deployment pipelines from minimal input information such as a raw training dataset. In particular, one example computing system can import a training dataset associated with a user. The computing system can execute an origination machine learning pipeline to perform a model architecture search that selects and trains a machine learning model for the training dataset. Execution of the origination machine learning pipeline can also result in generation of a deployment machine learning pipeline configured to enable deployment of the machine learning model (e.g., running the machine learning model to produce inferences and/or optionally other tasks such as re-training and/or re-tuning the model).

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model. A method includes: maintaining a plurality of training sessions; assigning, to each worker of one or more workers, a respective training session of the plurality of training sessions; repeatedly performing operations until meeting one or more termination criteria, the operations comprising: receiving an updated training session from a respective worker of the one or more workers, selecting a second training session, selecting, based on comparing the updated training session and the second training session using a fitness evaluation function, either the updated training session or the second training session as a parent training session, generating a child training session from the selected parent training session, and assigning the child training session to an available worker, and selecting a candidate model to be a trained model for the machine learning model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model. One of the methods includes receiving metadata for the training, generating a metadata sequence that represents the metadata, at each of a plurality of iterations: generating one or more trials that each specify a respective value for each of a set of hyperparameters, comprising, for each trial: generating an input sequence for the iteration that comprises (i) the metadata sequence and (ii) for any earlier trials, a respective sequence that represents the respective values for the hyperparameters specified by the earlier trial and a measure of performance for the trial, and processing an input sequence for the trial that comprises the input sequence for the iteration using a sequence generation neural network to generate an output sequence that represents respective values for the hyperparameters.

Abstract: An ion source having an extraction plate with a variable thickness is disclosed. The extraction plate has a protrusion on its interior or exterior surface proximate the extraction aperture. The protrusion increases the thickness of the extraction aperture in certain regions. This increases the loss area in those regions, which serves as a sink for ions and electrons. In this way, the plasma density is decreased more significantly in the regions where the extraction aperture has a greater thickness. The shape of the protrusion may be modified to achieve the desired plasma uniformity. Thus, it may be possible to create an extracted ion beam having a more uniform ion density. In some tests, the uniformity of the beam current along the width direction was improved by between 20% and 50%.

Abstract: An ion source including a chamber housing defining an ion source chamber and including an extraction plate on a front side thereof, the extraction plate having an extraction aperture formed therein, and a tubular cathode disposed within the ion source chamber and having an opening formed in a front half thereof nearest the extraction aperture, wherein a rear half of the tubular cathode furthest from the extraction aperture is closed.

Abstract: An ion source having an extraction plate with a variable thickness is disclosed. The extraction plate has a protrusion on its interior or exterior surface proximate the extraction aperture. The protrusion increases the thickness of the extraction aperture in certain regions. This increases the loss area in those regions, which serves as a sink for ions and electrons. In this way, the plasma density is decreased more significantly in the regions where the extraction aperture has a greater thickness. The shape of the protrusion may be modified to achieve the desired plasma uniformity. Thus, it may be possible to create an extracted ion beam having a more uniform ion density. In some tests, the uniformity of the beam current along the width direction was improved by between 20% and 50%.

Abstract: A hybrid superconductive device for stabilizing an electric grid comprises (a) a magnetic core arrangement at least partially carrying an AC winding the AC winding connectable to an AC circuit for a current to be limited in the event of a fault; (b) at least one superconductive coil configured for storing electromagnetic energy; the superconductive coil magnetically coupled with the core arrangement and saturating the magnetic core arrangement during use. The hybrid superconductive device further comprises a switch unit preprogrammed for switching electric current patterns corresponding to the following modes: at least partially charging the superconductive coil; a standby mode when the superconductive coil is looped back; and at least partially discharging the superconductive coil into the circuit. Optionally, hybrid superconductive device comprises at least one passage located within said magnetic flux. The passage conducts a material flow comprising components magnetically separable by said magnetic flux.

Abstract: A system for extending the life of a repeller in an IHC ion source is disclosed. The system includes an IHC ion source wherein the back surface of the repeller has been shaped to reduce the possibility of electrical shorts. The separation distance between the back surface of the repeller and the chamber wall behind the repeller is increased along its outer edge, as compared to the separation distance near the center of the repeller. This separation distance reduces the possibility that deposited material will flake and short the repeller to the chamber wall. Further, in certain embodiments, the separation distance between the back surface of the repeller and the chamber wall near the center of the repeller is unchanged, so as to minimize the flow of gas that exits from the chamber. The back surface of the repeller may be tapered, stepped or arced to achieve these criteria.

Abstract: An ion source including a chamber housing defining an ion source chamber and including an extraction plate on a front side thereof, the extraction plate having an extraction aperture formed therein, and a tubular cathode disposed within the ion source chamber and having an opening formed in a front half thereof nearest the extraction aperture, wherein a rear half of the tubular cathode furthest from the extraction aperture is closed.

Abstract: An ion source including a chamber housing defining an ion source chamber and including an extraction plate on a front side thereof, the extraction plate having an extraction aperture formed therein, and a tubular cathode disposed within the ion source chamber and having a slot formed in a front-facing semi-cylindrical portion thereof disposed in a confronting relationship with the extraction aperture, wherein a rear-facing semi-cylindrical portion of the tubular cathode directed away from the extraction aperture is closed.

Abstract: An ion source including a chamber housing defining an ion source chamber and including an extraction plate on a front side thereof, the extraction plate having an extraction aperture formed therein, and a tubular cathode disposed within the ion source chamber and having a slot formed in a front-facing semi-cylindrical portion thereof disposed in a confronting relationship with the extraction aperture, wherein a rear-facing semi-cylindrical portion of the tubular cathode directed away from the extraction aperture is closed.

Abstract: A hearing system comprising a hearing device and a connected device, wherein the hearing device comprises a microphone; a processor for processing a signal from the microphone; an output device for outputting the processed signal to an ear of the user; a memory for storing program code modules to be processed by the processor; and a transceiver for exchanging data with the connected device. In the memory of the hearing device a first set of program code modules and in the memory of the connected device a second set of program code modules different from the first set is stored. A program code module of the second set of program code modules is transferred from the connected device to the hearing device upon request from the hearing device or the connected device.

Abstract: Techniques for grammar-based automated generation of annotated synthetic form training data for machine learning are described. A training data generation engine utilizes a defined grammar to construct a layout for a form, select key-value units to place within the layout, and select attribute variants for the key-value units. The form is rendered and stored at a storage location, where it can be provided along with other similarly-generated forms to be used as training data for a machine learning model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model. A method includes: maintaining a plurality of training sessions; assigning, to each worker of one or more workers, a respective training session of the plurality of training sessions; repeatedly performing operations until meeting one or more termination criteria, the operations comprising: receiving an updated training session from a respective worker of the one or more workers, selecting a second training session, selecting, based on comparing the updated training session and the second training session using a fitness evaluation function, either the updated training session or the second training session as a parent training session, generating a child training session from the selected parent training session, and assigning the child training session to an available worker, and selecting a candidate model to be a trained model for the machine learning model.

Abstract: A hybrid superconductive device for stabilizing an electric grid comprises (a) a magnetic core arrangement at least partially carrying an AC winding the AC winding connectable to an AC circuit for a current to be limited in the event of a fault; (b) at least one superconductive coil configured for storing electromagnetic energy; the superconductive coil magnetically coupled with the core arrangement and saturating the magnetic core arrangement during use. The hybrid superconductive device further comprises a switch unit preprogrammed for switching electric current patterns corresponding to the following modes: at least partially charging the superconductive coil; a standby mode when the superconductive coil is looped back; and at least partially discharging the superconductive coil into the circuit. Optionally, hybrid superconductive device comprises at least one passage located within said magnetic flux. The passage conducts a material flow comprising components magnetically separable by said magnetic flux.

Abstract: An IHC ion source with multiple configurations is disclosed. For example, an IHC ion source comprises a chamber, having at least one electrically conductive wall, and a cathode and a repeller disposed on opposite ends of the chamber. Electrodes are disposed on one or more walls of the ion source. Bias voltages are applied to at least one of the cathode, repeller and the electrodes, relative to the electrically conductive wall of the chamber. Further, the IHC ion source comprises a configuration circuit, which receives the various voltages as input voltages, and provides selected output voltages to the cathode, repeller and electrodes, based on user input. In this way, the IHC ion source can be readily reconfigured for different applications without rewiring the power supplies, as is currently done. This configuration circuit may be utilized with other types of ion sources as well.

Abstract: Techniques for automated form understanding via layout-agnostic identification of keys and corresponding values are described. An embedding generator creates embeddings of pixels from an image including a representation of a form. The generated embeddings are similar for pixels within a same key-value unit, and far apart for pixels not in a same key-value unit. A weighted bipartite graph is constructed including a first set of nodes corresponding to keys of the form and a second set of nodes corresponding to values of the form. Weights for the edges are determined based on an analysis of distances between ones of the embeddings. The graph is partitioned according to a scheme to identify pairings between the first set of nodes and the second set of nodes that produces a minimum overall edge weight. The pairings indicate keys and values that are associated within the form.

Abstract: Techniques for layout-agnostic clustering-based classification of document keys and values are described. A key-value differentiation unit generates feature vectors corresponding to text elements of a form represented within an electronic image using a machine learning (ML) model. The ML model was trained utilizing a loss function that separates keys from values. The feature vectors are clustered into at least two clusters, and a cluster is determined to include either keys of the form or values of the form via identifying neighbors between feature vectors of the cluster(s) with labeled feature vectors.

Abstract: An indirectly heated cathode ion source having a cylindrical housing with two open ends is disclosed. The cathode and repeller are sized to fit within the two open ends. These components may be inserted into the open ends, creating a small radial spacing that provides electrical isolation between the cylindrical housing and the cathode and repeller. In another embodiment, the repeller may be disposed from the end of the cylindrical housing creating a small axial spacing. In another embodiment, insulators are used to hold the cathode and repeller in place. This design results in a reduced distance between the cathode column and the extraction aperture, which may be beneficial to the generation of ion beams of certain species.

Abstract: A hearing system comprising a hearing device and a connected device, wherein the hearing device comprises a microphone; a processor for processing a signal from the microphone; an output device for outputting the processed signal to an ear of the user; a memory for storing program code modules to be processed by the processor; and a transceiver for exchanging data with the connected device. In the memory of the hearing device a first set of program code modules and in the memory of the connected device a second set of program code modules different from the first set is stored. A program code module of the second set of program code modules is transferred from the connected device to the hearing device upon request from the hearing device or the connected device.