Abstract: A system for applying a neural network to an input instance. The neural network includes an optimization layer for determining values of one or more output neurons from values of one or more input neurons by a joint optimization parametrized by one or more parameters. An input instance is obtained. The values of the one or more input neurons to the optimization layer are obtained and input vectors for the one or more input neurons are determined therefrom. Output vectors for the one or more output neurons are computed from the determined input vectors by jointly optimizing at least the output vectors with respect to the input vectors to solve a semidefinite program defined by the one or more parameters. The values of the one or more output neurons are determined from the respective computed output vectors.

Abstract: In accordance with an embodiment, described herein are systems and methods for use with business process management (BPM) systems or environments, including software development tools for developing software applications that are associated with business processes (business process software applications, business process applications, process applications, or processes). In accordance with an embodiment, the system comprises a column-based process editor and graphical user interface, that can utilize a column-based format to model and display a process, for example a dynamic process, which can subsequently be deployed and executed within the system or environment. For example, the column-based process editor and graphical user interface can include various graphical elements to indicate configured properties of artifacts of a dynamic process. Interaction with the graphical elements enables users to review, edit, and/or remove particular properties associated with the dynamic process.

Abstract: A compiler receives a description of a machine learning network (MLN) and generates a computer program that implements the MLN on a machine learning accelerator (MLA). To implement the MLN, the compiler generates compute instructions that implement computations of the MLN on different processing units (Tiles), and data transfer instructions that transfer data used in the computations. The compiler may statically schedule at least a portion of the instructions for execution by the Tiles according to fixed timing. The compiler may initially implement data transfers between non-adjacent Tiles (or external memories) by implementing a sequence of transfers through one or more intermediate Tiles (or external memories) in accordance with a set of default routing rules that dictates the data path. The computer program may then be simulated to identify routing conflicts. When routing conflicts are detected, the compiler updates the computer program in a manner that avoids the conflicts.

Abstract: Dynamically overriding a function based on a capability set. A computer system reads a portion of an executable image file. The portion includes a first memory address corresponding to a first callee function implementation. The first memory address was inserted into the portion by a compiler toolchain. Based on extensible metadata included in the executable image file, and based on a capability set that is specific to the computer system, the computer system determines a second memory address corresponding to a second callee function implementation. Before execution of the portion, the computer system modifies the portion to replace the first memory address with the second memory address.

Abstract: Techniques that facilitate feedback loop learning between artificial intelligence systems are provided. In one example, a system includes a monitoring component and a machine learning component. The monitoring component identifies a data pattern associated with data for an artificial intelligence system. The machine learning component compares the data pattern to historical data patterns for the artificial intelligence system to facilitate modification of at least a component of the artificial intelligence system and/or one or more dependent systems of the artificial intelligence system.

Abstract: Code search is used to support various features. Code segments may be indexed using a code structure representation of the code segment. Code segments may be compared for similarity to identify a match with an input code segment using a comparison of logic trees generated for the input code segment and a stored code segment in an entry with a matching index value.

Abstract: Techniques for implementing a dynamic API bot for robotic process automation are disclosed. In some embodiments, a computer system performs operations comprising: providing a data file having a predefined template comprising dedicated fields for an identification of an API, a type of call method, metadata identifying one or more objects, and data of the object(s); providing a low-code no-code (LCNC) development platform configured to enable a user to develop a bot by dragging and dropping application components of the bot; receiving, via the LCNC development platform, a configuration of the bot comprising a configuration of the application components of the bot and an identification of the data file; and running the bot, the bot being configured to generate a request using the data file, converting the data of the object(s) into a payload in a format required by the API based on the data file.

Abstract: The disclosed computer-implemented method may include (1) receiving, at a hardware accelerator that supports an ANN, an activation data set that is to undergo a convolution operation via a filter kernel of the ANN, (2) receiving, at the hardware accelerator, an argument indicating that the filter kernel exceeds at least one boundary of the activation data set when slid across a certain position during the convolution operation, (3) determining, based at least in part on the argument, that the hardware accelerator is to generate padding data at the boundary of the activation data set in connection with the certain position of the filter kernel, and then (4) performing, at the hardware accelerator, the convolution operation by processing a portion of the activation data set and the padding data when the filter kernel slides across the certain position. Various other systems and methods are also disclosed.

Abstract: Embodiments described herein are generally directed to an edge-CaaS (eCaaS) framework for providing life-cycle management of containerized applications on the edge. According to an example, declarative intents are received indicative of a use case for which a cluster of a container orchestration platform is to be deployed within an edge site that is to be created based on infrastructure associated with a private network. A deployment template is created by performing intent translation on the declarative intents and based on a set of constraints. The deployment template identifies the container orchestration platform selected by the intent translation. The deployment template is then executed to deploy and configure the edge site, including provisioning and configuring the infrastructure, installing the container orchestration platform on the infrastructure, configuring the cluster within the container orchestration platform, and deploying a containerized application or portion thereof on the cluster.

Abstract: Using a natural language (NL) latent presentation in the automated conversion of source code from a base programming language (e.g., C++) to a target programming language (e.g., Python). A base-to-NL model can be used to generate an NL latent representation by processing a base source code snippet in the base programming language. Further, an NL-to-target model can be used to generate a target source code snippet in the target programming language (that is functionally equivalent to the base source code snippet), by processing the NL latent representation. In some implementations, output(s) from the NL-to-target model indicate canonical representation(s) of variables, and in generating the target source code snippet, technique(s) are used to match those canonical representation(s) to variable(s) of the base source code snippet. In some implementations, multiple candidate target source code snippets are generated, and a subset (e.g., one) is selected based on evaluation(s).

Abstract: A system for a heap dump conversion comprises a network storage device comprising a core dump file, a first server coupled to the network storage device, and a second server coupled to the network storage device and to the first server, where the second server receives a notification associated with the core dump file, obtains metadata information from the core dump file in response to receiving the notification, autonomously converts the core dump file into a heap dump file using the metadata information, and stores the heap dump file in the network storage device.

Abstract: A method of providing a control software configuration for a module of a modular plant, the method including, in a module engineering phase: receiving a user definition for the module; and automatically generating the control software configuration for the module based on the user definition, the automatically generating including: specifying parameters for the module that are not specific to any target system; and providing the control software configuration as a controller-agnostic configuration file for subsequent binding of controller software instantiated according to the controller-agnostic configuration file to a hardware controller of the module when the module is integrated into a target system during a plant engineering phase.

Abstract: A computation unit for performing a computation of a neural network layer is disclosed. A number of processing element (PE) units are arranged in an array. First input values are provided in parallel in an input dimension of the array during a first processing period, and a second input values are provided in parallel in the input dimension during a second processing period. Computations are performed by the PE units based on stored weight values. An adder coupled to the first set of PE units generates a first sum of results of the computations by the first set of PE units during the first processing cycle, and generates a second sum of results of the computations during the second processing cycle. A first accumulator coupled to the first adder stores the first sum, and further shifts the first sum to a second accumulator prior to storing the second sum.

Abstract: A computer-implemented method for carrying out an update management for updating application software on data processing units. The method includes: providing an error model, which indicates a number of errors across software versions of the application software; determining an update sequence for data processing units, depending on their particular software version; and updating the application software or providing an update prompt to update the application software in the data processing units according to the update sequence.

Abstract: The present disclosure provides a permission-controlled smart contract upgrade method. The method first deploys a contract upgrade smart contract, and then deploys a smart contract to be upgraded including information of a required minimum number of agreements on passing a proposal related to the contract. Then, any of the blockchain nodes receives a contract upgrade proposal submitted by one of the plurality of user clients. After the contract upgrade smart contract determines to pass the proposal, a proposing event is generated and then is forwarded to each user client participating in the contract. After the user client receives the proposing event, the user client receives user's vote and feeds it back to the contract upgrade smart contract. If the number of agreements exceeds the required minimum number of agreements, the proposal is passed and employed, and a binary replacement is performed to complete upgrade of the smart contract.

Abstract: In a vehicle master device, a first device includes an update data acquisition unit that is configured to acquire update data, an update data distribution unit that is configured to distribute the update data acquired by the update data acquisition unit to a rewrite target ECU, a process execution request unit that is configured to request the second device to execute at least a part of a process related to verification of the update data before the update data distribution unit distributes the update data, a processing result acquisition unit that is configured to acquire a processing result from the second device, and a verification unit that is configured to verify the update data using the processing result acquired by the processing result acquisition unit.

Abstract: The present disclosure relates generally to a system that includes a storage component that stores data and a processor. The processor may install an application using a portion of the storage component, partition the portion of the storage component from a remaining portion of the storage component, and execute the application via the portion of the storage component.

Abstract: Systems, machines, methods and products for generating a configured software solution using one or more configuration packages. A decision service may be configured to generate decision data based on a configuration package comprising user-generated input, a collection of configurations, and a decision flow template. The user-generated input may be used for selecting an artifact from an artifact library in a configuration database. The collection of configurations may be infused, dynamically, into the decision flow template. The decision flow template may be exposed for user modification. The decision flow template may be integrated into the configuration package in association with at least one configurable decision element and a user configuration selected from the collection of configurations for specifying one or more parameters in the artifact. The artifact and the user configuration may be combined with the decision flow template to generate the configured software solution.

Abstract: A method may include applying to at least a portion of historical runtimes associated with each of a plurality of tests included in a test suite, a machine learning model trained to identify one or more outlying runtime values. The portion of historical runtimes may include an n quantity of the most recent historical runtimes. An expected runtime for each test may be determined based on the portion of historical runtimes excluding the outlying runtime values. A schedule for executing each test in the test suite may be determined based on the expected runtime of each test. The test suite may be executed in accordance with the schedule. Moreover, the executing of the test suite may be monitored based on the expected runtime of each test. Related systems and computer program products are also provided.

Abstract: Methods for supporting a development of application source code using input from a first smart glasses of a first user and a second smart glasses of a second user is provided. Methods may include retrieving the application source code from an internal development platform and displaying the application source code on an augmented reality (“AR”) display of the first smart glasses and an AR display of the second smart glasses. Methods may include receiving, from the first smart glasses, a command to edit the application source code and in response, deactivating each input device of the second smart glasses. Methods may include receiving input of one or more edits on the first smart glasses and updating the application source code to include the input. Methods may include displaying the updated application source code on the AR displays and reactivating each input device of the second smart glasses.