Abstract: A system maintains a queue for storing packets, which are enqueued at a tail of the queue and dequeued at a head of the queue. The system computes a queue utilization value, based on the packets stored in the queue. The system computes an excess amount value, based on the packets stored in the queue and previously tagged as excess packets. The system receives a first packet at the tail of the queue and determines whether a difference between the queue utilization value and the excess amount value exceeds a predetermined threshold. Responsive to determining that the difference exceeds the predetermined threshold, the system tags the first packet as an excess packet. Responsive to tagging the first packet as an excess packet, the system performs an operation associated with the first packet or a second packet at the head of the queue to reduce congestion.

Abstract: A system maintains ordered sub-queues used for storing packets, which are to be dequeued by a scheduler. A respective is enqueued into a sub-queue, and a virtual time associated with the respective packet is based on a current sub-queue virtual time corresponding to a previously enqueued packet in the sub-queue. The system dequeues, by the scheduler, a first packet from a selected sub-queue and determines a packet virtual time associated with a next packet in the currently selected sub-queue. Responsive to determining that the packet virtual time associated with the next packet is greater than a current global virtual time, the system selects a next sub-queue in the ordered plurality of sub-queues. The system updates the current global virtual time based on a packet virtual time of the dequeued first packet.

Abstract: Systems and methods are provided for optimally allocating resources used to perform multiple tasks/jobs, e.g., machine learning training jobs. The possible resource configurations or candidates that can be used to perform such jobs are generated. A first batch of training jobs can be randomly selected and run using one of the possible resource configuration candidates. Subsequent batches of training jobs may be performed using other resource configuration candidates that have been selected using an optimization process, e.g., Bayesian optimization. Upon reaching a stopping criterion, the resource configuration resulting in a desired optimization metric, e.g., fastest job completion time can be selected and used to execute the remaining training jobs.

Abstract: Systems and methods are provided for updating resource allocation in a distributed network. For example, the method may comprise allocating a plurality of resource containers in a distributed network in accordance with a first distributed resource configuration. Upon determining that a processing workload value exceeds a stabilization threshold of the distributed network, determining a resource efficiency value of the plurality of resource containers in the distributed network. When a resource efficiency value is greater than or equal to the threshold resource efficiency value, the method may generate a second distributed resource configuration that includes a resource upscaling process, or when the resource efficiency value is less than the threshold resource efficiency value, the method may generate the second distributed resource configuration that includes a resource outscaling process. The resource allocation may transmit the second to update the resource allocation.

Abstract: Systems, methods, and computer-readable media are described for employing a machine learning-based approach such as adaptive Bayesian optimization to learn over time the most optimized assignments of incoming network requests to service function chains (SFCs) created within network slices of a 5G network. An optimized SFC assignment may be an assignment that minimizes an unknown objective function for a given set of incoming network service requests. For example, an optimized SFC assignment may be one that minimizes request response time or one that maximizes throughput for one or more network service requests corresponding to one or more network service types. The optimized SFC for a network request of a given network service type may change over time based on the dynamic nature of network performance. The machine-learning based approaches described herein train a model to dynamically determine optimized SFC assignments based on the dynamically changing network conditions.

Abstract: In some examples, a system including physical graphics processing units (GPUs) receives a request to schedule a new job to be executed in the system that is accessible by a plurality of tenants to use the physical GPUs. The system allocates the new job to a collection of vGPUs of the physical GPUs based on an operational cost reduction objective to reduce a cost associated with a usage of the physical GPUs and based on a tenant isolation constraint to provide tenant isolation wherein a single tenant of the plurality of tenants is to use a physical GPU at a time.

Abstract: In hemodynamic determination in medical imaging, the classifier is trained from synthetic data rather than relying on training data from other patients. A computer model (in silico) may be perturbed in many different ways to generate many different examples. The flow is calculated for each resulting example. A bench model (in vitro) may similarly be altered in many different ways. The flow is measured for each resulting example. The machine-learnt classifier uses features from medical scan data for a particular patient to estimate the blood flow based on mapping of features to flow learned from the synthetic data. Perturbations or alterations may account for therapy so that the machine-trained classifier may estimate the results of therapeutically altering a patient-specific input feature. Uncertainty may be handled by training the classifier to predict a distribution of possibilities given uncertain input distribution.

Abstract: A device and corresponding method are provided determining a consumed computing capacity of a first networking device exceeds the threshold for total capacity for processing monitoring data for a monitoring metric. An optimization engine determines a second networking device with unused computing capacity sufficient for processing the monitoring data generated by the first networking device. The optimization engine automatically moves the monitoring data for the monitoring metric generated by the first networking device to the second networking device and causes the second networking device to process the monitoring data.

Abstract: A medical knowledge base in a digital, clinical system is upgraded. A storage with a knowledge base, being a SNOMED knowledge base, is provided in a web ontology format. Procedural data, representing clinical procedures for evaluation of a patient's health state, is received. The received procedural data is mapped in a set of SNOMED expressions. The SNOMED expressions are converted into statements in the web ontology format. The SNOMED knowledge base is upgraded with the received procedural data by adding the statements in the SNOMED knowledge base for providing a processable file with an upgraded version of the SNOMED knowledge base.

Abstract: A technique for determining a cardiac metric from rest and stress perfusion cardiac magnetic resonance (CMR) images is provided. A neural network system for determining at least one cardiac metric from CMR images comprises an input layer configured to receive at least one CMR image representative of a rest perfusion state and at least one CMR image representative of a stress perfusion state. The neural network system further comprises an output layer configured to output at least one cardiac metric based on the at least one CMR image representative of the rest perfusion state and the at least one CMR image representative of the stress perfusion state. The neural network system with interconnections between the input layer and the output layer is trained by a plurality of datasets.

Abstract: Systems and methods for determining corresponding locations of points of interest in a plurality of input medical images are provided. A plurality of input medical images comprising a first input medical image and one or more additional input medical images is received. The first input medical image identifies a location of a point of interest. A set of features is extracted from each of the plurality of input medical images. Features between each of the sets of features are related using a machine learning based relational network. A location of the point of interest in each of the one or more additional input medical images that corresponds to the location of the point of interest in the first input medical image is identified based on the related features. The location of the point of interest in each of the one or more additional input medical images is output.

Abstract: An encoder is trained together with a perception component based on a training set comprising unannotated sensor data sets and annotated sensor data sets in a sequence of multiple training steps.

Abstract: Systems and methods for automated assessment of a vessel are provided. One or more input medical images of a vessel of a patient are received. A plurality of vessel assessment tasks for assessing the vessel is performed using a machine learning based model trained using multi-task learning. The plurality of vessel assessment tasks comprises segmentation of reference walls of the vessel from the one or more input medical images and segmentation of lumen of the vessel from the one or more input medical images. Results of the plurality of vessel assessment tasks are output.

Abstract: Systems and methods for determining a quantity of interest of a patient comprise receiving patient data of the patient at a first physiological state. A value of a quantity of interest of the patient at the first physiological state is determined based on the patient data. The quantity of interest represents a medical characteristic of the patient. Features are extracted from the patient data, wherein the features which are extracted are based on the quantity of interest to be determined for the patient at a second physiological state. The value of the quantity of interest of the patient at the first physiological state is mapped to a value of the quantity of interest of the patient at the second physiological state based on the extracted features.

Abstract: Systems and methods are provided for optimizing a serverless workflow. Given a directed acyclic graph (“DAG”) defining functional relationships and a gamma tuning factor to indicate a preference between cost and performance, a serverless workflow corresponding to the DAG may be optimized. The optimization is carried out in accordance with the gamma tuning factor, and is carried out in sub-segments of the DAG called stages. In addition, systems for allowing disparate types of storage media to be utilized by a serverless platform to store data are disclosed. The serverless platforms maintain visibility of the storage media types underlying persistent volumes, and may store data in partitions across disparate types of storage media. For instance, one item of data may be stored partially at a byte addressed storage media and partially at a block addressed storage media.

Abstract: Systems and methods can be configured to determine a plurality of computing resource configurations used to perform machine learning model training jobs. A computing resource configuration can comprise: a first tuple including numbers of worker nodes and parameter server nodes, and a second tuple including resource allocations for the worker nodes and parameter server nodes. At least one machine learning training job can be executed using a first computing resource configuration having a first set of values associated with the first tuple. During the executing the machine learning training job: resource usage of the worker nodes and parameter server nodes caused by a second set of values associated with the second tuple can be monitored, and whether to adjust the second set of values can be determined. Whether a stopping criterion is satisfied can be determined. One of the plurality of computing resource configurations can be selected.

Abstract: Systems and methods are provided for improving autotuning procedures. For example, the system can implement a task launcher, a scheduler, and an agent to launch, schedule, and execute decomposed autotuning stages, respectively. The scheduling policy implemented by the scheduler may perform operations beyond a simple scheduling policy (e.g., a FIFO-based scheduling policy), which produces a high queuing delay. By leveraging autotuning specific domain knowledge, this may help reduce queuing delay and improve resource utilization that is otherwise found in traditional systems.

Abstract: For decision support based on perfusion in medical imaging, a machine-learned model, such as a model trained with deep learning, generates perfusion examination information from CT scans of the patient. Other information, such as patient-specific information, may be used with the CT scans to generate the perfusion examination information. Since a machine-learned model is used, the perfusion examination information may be estimated from a spatial and/or temporally sparse number of scan shots or amount of CT dose. The results of perfusion imaging may be provided with less than the current, standard, or typical radiation dose.

Abstract: Systems and methods are provided for improved TCP congestion control designed to address “mixed coarse-grained-fine-grained signal” scenarios. A TCP sender of the present technology achieves this improvement by leveraging two TCP congestion windows for a TCP connection: (1) a “fine-grained TCP signal-dependent congestion window” which is adjusted in response to “fine-grained” TCP congestion signals (as intelligently classified/defined by the present technology); and (2) a “coarse-grained TCP signal-dependent congestion window” which is adjusted in response to “coarse-grained” TCP congestion signals (as intelligently classified/defined by the present technology). With these two novel/unique congestion windows at disposal, the TCP sender can then dynamically (and intelligently) select an appropriate congestion window for dictating packet transmission for a TCP connection (e.g., the contemporaneously smaller congestion window).

Abstract: Techniques for processing multiple cardiac images are disclosed. The processing may take place either during or after an angiography exam of a coronary artery of interest. The multiple cardiac images are obtained either during or after the angiography exam. Each of the multiple cardiac images depicts a respective segment of the coronary artery of interest. A geometric structure of the coronary artery of interest is determined based on the multiple cardiac images. A lumped parameter model of the coronary artery of interest is determined based on the geometric structure, and respective values of at least one hemodynamic index at a position of the coronary artery of interest is determined based on the lumped parameter model of the coronary artery of interest.