Abstract: Techniques are described for performing active surveillance and learning for machine learning (ML) model authoring and deployment workflows. In an embodiment, a method comprises applying, by a system comprising a processor, a primary ML model trained on a training dataset to data samples excluded from the training dataset to generate inferences based on the data samples. The method further comprises employing, by the system, one or more active surveillance techniques to regulate performance of the primary ML model in association with the applying, wherein the one or more active surveillance techniques comprise at least one of, performing a model scope evaluation of the primary ML model relative to the data samples or using a domain adapted version of the primary ML model to generate the inferences.

Abstract: Systems and techniques that facilitate synthetic training data generation for improved machine learning generalizability are provided. In various embodiments, an element augmentation component can generate a set of preliminary annotated training images based on an annotated source image. In various aspects, a preliminary annotated training image can be formed by inserting at least one element of interest or at least one background element into the annotated source image. In various instances, a modality augmentation component can generate a set of intermediate annotated training images based on the set of preliminary annotated training images. In various cases, an intermediate annotated training image can be formed by varying at least one modality-based characteristic of a preliminary annotated training image. In various aspects, a geometry augmentation component can generate a set of deployable annotated training images based on the set of intermediate annotated training images.

Abstract: Various methods and systems are provided for automatically registering and stitching images. In one example, a method includes entering a first image of a subject and a second image of the subject to a model trained to output a transformation matrix based on the first image and the second image, where the model is trained with a plurality of training data sets, each training data set including a pair of images, a mask indicating a region of interest (ROI), and associated ground truth, automatically stitching together the first image and the second image based on the transformation matrix to form a stitched image, and outputting the stitched image for display on a display device and/or storing the stitched image in memory.

Abstract: Systems/techniques that facilitate learning-based clean data selection are provided. In various embodiments, a system can access a raw dataset. In various aspects, the system can select, via execution of a data selection machine learning model, a clean dataset from the raw dataset. In various instances, the system can train a target machine learning model to perform a target task based on the clean dataset. In various aspects, the clean dataset can include candidate-annotation groupings that are in the raw dataset and that are determined by the data selection machine learning model to be suitable for training of the target machine learning model, and the clean dataset can exclude candidate-annotation groupings that are in the raw dataset and that are determined by the data selection machine learning model to not be suitable for training of the target machine learning model.

Abstract: A computer-implemented system is provided that includes a learning network component that determines respective weights assigned to respective node inputs of the learning network in accordance with a learning phase of the learning network and trains a variable separator component to differentially change learning rates of the learning network component. A differential rate component applies at least one update learning rate to adjust at least one weight assigned to at least one of the respective node inputs and applies at least one other update learning rate to adjust the respective weight assigned to at least one other of the respective node inputs in accordance with the variable separator component during the learning phase of the learning network.

Abstract: Techniques regarding autonomous data augmentation are provided. For example, one or more embodiments described herein can regard a system comprising a memory that can store computer-executable components. The system can also comprise a processor, operably coupled to the memory, that executes the computer-executable components stored in the memory. The computer-executable components can include a data augmentation component that executes a random unidimensional augmentation algorithm to augment a dataset for training a machine learning model via a plurality of augmentation operations. The random unidimensional augmentation algorithm can employ a global augmentation parameter that defines: a distortion magnitude associated with the plurality of augmentation operations, and a number of augmentation operations included in the plurality of augmentation operations.

Abstract: Techniques are provided for compressing deep neural networks using a structured filter pruning method that is extensible and effective. According to an embodiment, a computer-implemented method comprises determining, by a system operatively coupled to a processor, importance scores for filters of layers of a neural network model previously trained until convergence for an inferencing task on a training dataset. The method further comprises removing, by the system, a subset of the filters from one or more layers of the layers based on the importance scores associated with the subset failing to satisfy a threshold importance score value. The method further comprises converting, by the system, the neural network model into a compressed neural network model with the subset of the filters removed.

Abstract: Systems and techniques that facilitate synthetic training data generation for improved machine learning generalizability are provided. In various embodiments, an element augmentation component can generate a set of preliminary annotated training images based on an annotated source image. In various aspects, a preliminary annotated training image can be formed by inserting at least one element of interest or at least one background element into the annotated source image. In various instances, a modality augmentation component can generate a set of intermediate annotated training images based on the set of preliminary annotated training images. In various cases, an intermediate annotated training image can be formed by varying at least one modality-based characteristic of a preliminary annotated training image. In various aspects, a geometry augmentation component can generate a set of deployable annotated training images based on the set of intermediate annotated training images.

Abstract: A computer-implemented system is provided that includes a learning network component that determines respective weights assigned to respective node inputs of the learning network in accordance with a learning phase of the learning network and trains a variable separator component to differentially change learning rates of the learning network component. A differential rate component applies at least one update learning rate to adjust at least one weight assigned to at least one of the respective node inputs and applies at least one other update learning rate to adjust the respective weight assigned to at least one other of the respective node inputs in accordance with the variable separator component during the learning phase of the learning network.

Abstract: A tenant model models workload usage of tenants, based upon a set of tenant attributes. The model is applied to a set of tenants waiting to be on-boarded to a workload to identify a metric indicative of likely tenant usage of the workload. A subset, of the set of tenants, are identified for on-boarding, based upon the metric, and on-boarding functionality is controlled to the identified subset of tenants.

Abstract: Techniques are described for performing active surveillance and learning for machine learning (ML) model authoring and deployment workflows. In an embodiment, a method comprises applying, by a system comprising a processor, a primary ML model trained on a training dataset to data samples excluded from the training dataset to generate inferences based on the data samples. The method further comprises employing, by the system, one or more active surveillance techniques to regulate performance of the primary ML model in association with the applying, wherein the one or more active surveillance techniques comprise at least one of, performing a model scope evaluation of the primary ML model relative to the data samples or using a domain adapted version of the primary ML model to generate the inferences.

Abstract: Systems, apparatus, instructions, and methods for medical machine time-series event data processing are disclosed. An example apparatus includes a data processor to process one-dimensional data captured over time with respect to patient(s). The example apparatus includes a visualization processor to transform the processed data into graphical representations and to cluster the graphical representations including the first graphical representation into at least first and second blocks arranged with respect to an indicator of a criterion to provide a visual comparison of the first block and the second block with respect to the criterion. The example apparatus includes an interaction processor to facilitate interaction, via the graphical user interface, with the first and second blocks of graphical representations to extract a data set for processing from at least a subset of the first and second blocks.

Abstract: Systems, apparatus, instructions, and methods for medical machine time-series event data generation are disclosed. An example synthetic time series data generation apparatus is to generate a synthetic data set including multi-channel time-series data and associated annotation using a first artificial intelligence network model. The example apparatus is to analyze the synthetic data set with respect to a real data set using a second artificial intelligence network model. When the second artificial intelligence network model classifies the synthetic data set as a first classification, the example apparatus is to adjust the first artificial intelligence network model using feedback from the second artificial intelligence network model. When the second artificial intelligence network model classifies the synthetic data set as a second classification, the example apparatus is to output the synthetic data set.

Abstract: An artificial intelligence platform and associated methods of training and use are disclosed. An example apparatus includes a data pipeline to: preprocess data using one or more preprocessing operations applied to features associated with the data; and enable debugging to visualize the preprocessed data. The example apparatus includes a network to: instantiate one or more differentiable operations in a training configuration to train an artificial intelligence model; capture feedback including optimization and loss information to adjust the training configuration; and store one or more metrics to evaluate performance of the artificial intelligence model.

Abstract: Systems and computer-implemented methods for facilitating automated compression of artificial neural networks using an iterative hybrid reinforcement learning approach are provided. In various embodiments, a compression architecture can receive as input an original neural network to be compressed. The architecture can perform one or more compression actions to compress the original neural network into a compressed neural network. The architecture can then generate a reward signal quantifying how well the original neural network was compressed. In (?)-proportion of compression iterations/episodes, where ??[0,1], the reward signal can be computed in model-free fashion based on a compression ratio and accuracy ratio of the compressed neural network. In (1??)-proportion of compression iterations/episodes, the reward signal can be predicted in model-based fashion using a compression model learned/trained on the reward signals computed in model-free fashion.

Abstract: A tenant model models workload usage of tenants, based upon a set of tenant attributes. The model is applied to a set of tenants waiting to be on-boarded to a workload to identify a metric indicative of likely tenant usage of the workload. A subset, of the set of tenants, are identified for on-boarding, based upon the metric, and on-boarding functionality is controlled to the identified subset of tenants.

Abstract: A tenant model models workload usage of tenants, based upon a set of tenant attributes. The model is applied to a set of tenants waiting to be on-boarded to a workload to identify a metric indicative of likely tenant usage of the workload. A subset, of the set of tenants, are identified for on-boarding, based upon the metric, and on-boarding functionality is controlled to the identified subset of tenants.

Abstract: A tenant model models workload usage of tenants, based upon a set of tenant attributes. The model is applied to a set of tenants waiting to be on-boarded to a workload to identify a metric indicative of likely tenant usage of the workload. A subset, of the set of tenants, are identified for on-boarding, based upon the metric, and on-boarding functionality is controlled to the identified subset of tenants.