Abstract: The present disclosure provides for integrated cooling systems including backside power delivery and methods of manufacturing the same. An integrated cooling assembly may include a device and a cold plate. The cold plate has a first side and an opposite second side, the first side having a recessed surface, sidewalls around the recessed surface that extend downwardly therefrom to define a cavity, and a plurality of support features disposed in the cavity. The first side of the cold plate is attached to a backside of the device to define a coolant channel therebetween. The cold plate includes a substrate, a dielectric layer disposed on a first surface of the substrate, a first conductive layer disposed between the first surface and the dielectric layer, a second conductive layer disposed on a second surface of the substrate, and thru-substrate interconnects connecting the first conductive layer to the second conductive layer.

Abstract: A computer-implemented method for identifying a problem list section from an electronic document includes receiving, by one or more processors, the electronic document, generating, by the one or more processors and based on applying an optical character recognition algorithm to the electronic document, unstructured text, and identifying, by the one or more processors, one or more problem list words in the unstructured text, the one or more problem list words belonging in a dataset for identifying a presence of a problem list section. The method also includes associating, by the one or more processors, a portion of the unstructured text that corresponds to the one or more problem list words in the unstructured text with the problem list section and outputting, by the one or more processors, at least a portion of the problem list section.

Abstract: Techniques for reducing power consumption in a radio unit are disclosed. The techniques include determining a number of Multiple-Input-Multiple-Output (MIMO) antenna elements of a MIMO antenna panel required to serve one or more User Equipment (UEs), identifying a number of MIMO antenna elements of the MIMO antenna panel not needed to serve the UEs. The techniques include sending, from an Open Radio Network (O-RAN) Distributed Unit (O-DU) to a O-RAN Radio Unit (O-RU), a beamforming weight of zero assigned to the MIMO antenna elements identified as not needed to serve the one or more UEs, the beamforming weight being sent in a digital domain via a Control plane (C-plane). The MIMO antenna elements assigned the beamforming weight of zero may be disabled thereby reducing the power consumption of the O-RU. Artificial Intelligence algorithms may be implemented to dynamically determine areas where the O-RAN should have more coverage or less coverage.

Abstract: A method (400) is provided for assisting in a configuring of charging triggers in a Network Function (410), NF. The method is performed by a Converged Charging Function (120), CHF. The method includes sending a notification message (420) from the CHF to the NF, wherein the notification message at least indicates one or more charging triggers to be configured by the NF in accordance with the notification message. A corresponding method of configuring charging triggers in an NF includes receiving the notification message and configuring the charging triggers in accordance therewith. Corresponding entities, computer programs and computer program products are also provided.

Abstract: A method and a system for automated performance of capacity allocation, brokerage, placement, and provisioning of compute, network, and storage resources are provided. The method includes: receiving a first data set that relates to resource requirements of a user; retrieving, from a memory, a second data set that relates to resource availability; analyzing the first data set and the second data set in order to determine a proposed allocation of resources and a proposed timing that corresponds to the proposed allocation; and provisioning the resources to the user based on the proposed allocation and the proposed timing. A machine learning model that is trained by using historical resource allocation data may be applied to the first data set and the second data set in order to perform the analysis.

Abstract: Microelectronic integrated circuit package structures include a die having a dielectric die edge sidewall and a bulk silicon die edge sidewall, where the bulk silicon die edge sidewall is in substantial alignment with the dielectric die edge sidewall. The bulk silicon die edge sidewall has a plurality of scallop structures along a vertical distance of the bulk silicon die edge sidewall.

Abstract: Embodiments herein provide for device packages comprising an integrated cooling assembly and methods of cooling packaged devices. The integrated cooling assembly comprising a semiconductor device, a manifold attached to the semiconductor device, and a sonic transducer attached to the manifold. The manifold comprises a top portion and a waveguide extending downwardly from the top portion. The sonic transducer is attached to the top portion. The top portion, the waveguide, and a backside of the semiconductor device collectively define a coolant chamber volume therebetween.

Abstract: A system comprises at least one camera disposed in an environment, and at least one computing node which hosts and executes an energy management system. The energy management system is configured to utilize computer vision processing of the images of the environment to extract information from the images, to utilize the extracted information from the images to determine usage patterns of light sources in the environment, and to generate intelligent recommendations for automated control of the light sources in the environment based on the determined usage patterns, to conserve energy consumption from use of the light sources in the environment.

Abstract: A turbine engine includes a rotor, a stator, and a seal assembly disposed between the rotor and the stator. The seal assembly includes seal segment. The seal segment includes a seal face that is configured to form a fluid bearing with the rotor. A lift channel extends within the seal segment from an opening on the seal face. The turbine engine further includes a spring assembly disposed within the lift channel. The spring assembly including a biasing element and a piston element coupled to the biasing element. The lift channel includes a lift volume portion extending is between the opening and the piston element. The piston element is movable within the lift channel based on a pressure within the fluid bearing to adjust a size of the lift volume portion.

Abstract: An example method includes storing status data for a network communication device connected to a host device by a pluggable connector in a memory of the network communication device. A change in the status data is signaled using a signal on the pluggable connector. The status data is sent from the memory of the network communication device to the host device over the pluggable connector. The status data is displayed on a local indicator of the host device.

Abstract: A method for training a machine learning model using positive and negative synthetic data is implemented via a computing system including a processor. The method includes generating synthetic data using a generative pre-trained transformer bidirectional language model and self-supervising the generated synthetic data based on positive traits including rule-based criteria and/or model-based criteria. The method also includes generating a set of positive synthetic data labels with gradient scale rating based on the self-supervised synthetic data, synthesizing a set of negative synthetic data labels by self-supervising the positive synthetic data labels, and training a machine learning model using the set of positive synthetic data labels and the set of negative synthetic data labels. Another method further includes utilizing the trained machine learning model to generate Objectives and Key Results (OKRs) within the context of an enterprise application.

Abstract: Embodiments described herein relate to dynamic selection of radio access technologies (RATs) available for use by 5G wireless devices. Managing selection of RATs to use by a 5G (and 4G LTE) capable wireless device can improve performance, e.g., when an application requires a higher throughput and a 5G SA connection underperforms a 5G NSA connection or a 4G LTE connection. Wireless devices can provide geolocation tagged cellular wireless network information, e.g., network performance measurements and observed properties, to network-based servers that collate and organize the crowd-sourced cellular wireless network information into a location database. The wireless device can use the cellular wireless network information of the location database for a geolocation at which the wireless device is operating to estimate performance for alternate wireless connections and compare to measured performance of a currently used wireless connection to determine whether to switch wireless RAT connections.

Abstract: In one example, a non-transitory computer-readable storage medium with program instructions for determining anomalies in machine learning models and impact on business output data thereof is disclosed. The method accesses machine learning model variables for a machine learning model, and determines a deviation in behavior of the machine learning model by performing statistical analysis on each machine learning model variable to identify statistical differences between machine learning model variable data captured in a given time window and data captured in a previous time window at the same time of a prior day, prior week, etc. If statistical differences exist, the method performs anomaly detection to determine whether the statistical differences are an anomaly. If so, the method determines whether the anomaly causes an impact on business decision output data, and if so, identifying a root cause and generating a notification alert for corrective action.

Abstract: State of the art approaches for training chained neural network models for delay prediction train the data models using only real data and not predicted data. Such models when used in a chained way leads to worse results as they are not exposed to predicted data during training. This leads to the model prediction errors showing sharp increase as the models tries to predict for subsequent stations past the immediate station. Embodiments disclosed herein provide a method and system for training of chained neural networks for delay prediction in transit networks. In this approach, a chained neural network model used by the system is trained such that data containing a mix of real data and predicted data is used for training each data model in a sequence of data models in the chained neural network model.

Abstract: There is a large latency and controller bandwidth associated with moving data between dies or between memory devices. The controller includes one or more flash interface modules (FIMs) that are utilized to write data to the memory device and read data from the memory device. Each of the one or more FIMs includes one or more switches. Each switch is utilized to transfer data from a source block to a destination block. Likewise, rather than using a memory external to the FIM to cache the data, the data is stored in a FIM cache and moved from the FIM cache to the relevant physical layer to be programmed to the destination block. Because data is not being transferred to the system memory, the latency and bandwidth associated with relocating data may be decreased.

Abstract: Described herein is a network and renderer aware split rendering technique that enables a cloud gaming service to split execution of cloud gaming workloads between a graphics processor of a cloud gaming server and a graphics processor of a client of the cloud gaming service. The split rendering technique enables portions of a frame that most vulnerable to quality degradation at low bitrates to be rendered on the client, which results in an improvement of the quality of the final frame that is presented at the client device.

Abstract: An additive manufacturing system includes a control system communicatively coupled to a consolidation device and configured to control operation of the consolidation device. The control system is configured to generate a model of a component including a plurality of elements and at least one region of interest. The control system is also configured to apply at least one strain load to at least one element of the plurality of elements and generate a build characteristic contribution profile based on the at least one strain load. The control system is further configured to determine a build parameter based at least partly on the build characteristic contribution profile.

Abstract: Disclosed are compounds of Formula (I) or a salt thereof, wherein R1 is: each W is independently NR1b or O; Z is a bond or CHR1d; and R1, R2, Rd, R3a, R3b, L1, B, V, Y, and n are defined herein. Also disclosed are methods of using such compounds as inhibitors of ROMK, and pharmaceutical compositions comprising such compounds. These compounds are useful in treating cardiovascular diseases.

Abstract: Integrated circuits and method of making including a plurality of transistors, each of the transistors including a source, a drain, and a channel located between the source and the drain. The integrated circuit also includes a plurality of buried power rails including a plurality of VSS power rails and a plurality of VDD power rails, at least one VSS pad and at least one VDD pad, a plurality of vias electrically connecting the at least one VSS pad to at least two of the plurality of VSS power rails, and a plurality of vias electrically connecting the at least one VDD pad to at least two of the plurality of buried VDD power rails.

Abstract: A system for facilitating efficient multicast traffic distribution in a network comprising a plurality of gateway switches is provided. During operation, the system can receive a join request for a first multicast group. The system can select, based on a source selection mechanism, a first datacenter from a plurality of datacenters as a source of a first multicast flow of the first multicast group. Here, a respective datacenter can provide the first multicast flow. The system can also select, based on a receiver selection mechanism, a first gateway switch from the plurality of gateway switches for receiving the first multicast flow. The system can then send, to the remote gateway switches, a control message comprising information indicating the selection of the source and receiver of the first multicast flow. The control message can notify the first gateway switch to request the first flow from the first datacenter.