Abstract: A method to detect anomalous behavior in a computing system begins by training a graph neural network (GNN) in an unsupervised manner by applying contrastive representation learning on sets of positive samples and negative samples derived from one or more heterogeneous graphs using meta-path sampling. Following training, a temporal graph derived from system-generated events is received. The GNN is used to embed the temporal graph into a vector representation in a vector space. The trained GNN is also used to embed a set of attack pattern graphs into corresponding vector representations in the vector space. For anomaly detection, the representation corresponding to the temporal graph is compared to the representations corresponding to the attack pattern graphs. In one embodiment, the comparison is implemented using a fuzzy pattern matching algorithm. If a fuzzy match is found, an indication that the temporal graph is associated with a potential attack on the computing system is then output.

Abstract: A natural language input is communicated to a back end application. Candidate agent data communicated from the back end application is presented. The candidate agent data comprises data of a set of agents considered by the back end application for performance of a command extracted from the natural language input. Candidate agent score data communicated from the back end application is presented. The candidate agent score data comprises score data of an agent in the set of agents. Selected agent data communicated from the back end application is presented. The selected agent data comprises data of a selected agent in the set of agents, the selected agent selected by the back end application for performance of the command. Skill performance data communicated from the back end application is presented. The skill performance data comprises data of a processing step performed by the selected agent.

Abstract: A computer-implemented method is provided for generating a language model for an application. The method includes estimating interpolation weights of each of a plurality of language models according to an Expectation Maximization (EM) algorithm based on a first metric. The method further includes classifying the plurality of language models into two or more sets based on characteristics of the two or more sets. The method also includes estimating a hyper interpolation weight for the two or more sets based on a second metric specific to the application. The method additionally includes interpolating the plurality of language models using the interpolation weights and the hyper interpolation weight to generate a final language model.

Abstract: System, apparatus, and/or a method for checking hardware components in an information handling system is disclosed that includes selecting a hardware component for testing; supplying power to the hardware component independent from the power supply provided by the information handling system; performing testing on the selected hardware component using power supplied by the independent power supply; determining the success or failure of the testing on the selected hardware component; indicating the result of the testing on the selected hardware component; and if the result of the testing on the selected hardware component is successful indicate the hardware component passed the testing, and if the result of the testing on the selected hardware component is not successful re-seat the hardware component and perform the testing on the selected hardware again.

Abstract: A method, apparatus and computer program product to detect whether specific sensitive data of a client is present in a cloud computing infrastructure is implemented without requiring that data be shared with the cloud provider, or that the cloud provider provide the client access to all data in the cloud. Instead of requiring the client to share its database of sensitive information, preferably the client executes a tool that uses a cryptographic protocol, namely, Private Set Intersection (PSI), to enable the client to detect whether their sensitive information is present on the cloud. Any such information identified by the tool is then used to label a document or utterance, send an alert, and/or redact or tokenize the sensitive data.

Abstract: In an embodiment, a method includes measuring a first number of control qubits in a quantum algorithm, wherein a quantum circuit representation of the quantum algorithm includes a multiple-controlled-NOT gate. In an embodiment, a method includes measuring a second number of ancilla qubits in a quantum computer. In an embodiment, a method includes comparing the first number and the second number to determine an optimum compilation method for a quantum circuit. In an embodiment, a method includes compiling, in response to the comparison determining the second number is greater than one and less than the difference of the first number and 2, a quantum circuit from the quantum algorithm using a hybrid method.

Abstract: A student neural network may be trained by a computer-implemented method, including: selecting a teacher neural network among a plurality of teacher neural networks, inputting an input data to the selected teacher neural network to obtain a soft label output generated by the selected teacher neural network, and training a student neural network with at least the input data and the soft label output from the selected teacher neural network.

Abstract: Selecting tables for compression by threshold statistical values. Identified tables are reordered according to fields having the lowest cardinality to increase the size of character strings replaced by keys during compression. Field locations are mapped between the original table and the reordered table. Dictionary-based compression is performed on reordered tables.

Abstract: Members of an audience of a visually dynamic event are clustered using a plurality of sources. A current point of interest (POI) of the visually dynamic event and a future POI of the visually dynamic event are identified across the member clusters. An effectiveness score for given content and a given member cluster is computed for the current point of interest and an effectiveness score for a given content and a given member cluster is computed for the future POI by tracking a position, a speed, and a direction of movement of the current POI. A location of a background area is determined and ranked for each member cluster based on the current point of interest. Electronic displays that correspond to the ranked background areas are identified and ranked. Content is distributed across the electronic displays for each current time period based on the ranked display areas.

Abstract: A neuromorphic device includes a plurality of first control lines, a plurality of second control lines and a matrix of resistive processing unit cells. Each resistive processing unit cell is electrically connected with one of the first control lines and one of the second control lines. A given resistive processing unit cell includes a first resistive device and a second resistive device. The first resistive device is a positively weighted resistive device and the second resistive device is a negatively weighted resistive device.

Abstract: Embodiments relate to a system, program product, and method for implementing loop lock reservations, and, more specifically, for holding a lock reservation across some or all of the iterations of a loop, and under certain conditions, temporarily effect a running thread to yield the reservation and allow other threads to enter the lock.

Abstract: A non-volatile memory cell includes a thin film resistor (TFR) in series and between a top state influencing electrode and a top wire. The TFR limits or generally reduces the electrical current at the top state influencing electrode from the top wire. As such, non-volatile memory cell endurance may be improved and adverse impacts to component(s) that neighbor the non-volatile memory cell may be limited. The TFR is additionally utilized as an etch stop when forming a top wire trench associated with the fabrication of the top wire. In some non-volatile memory cells where cell symmetry is desired, an additional TFR may be formed between a bottom wire and a bottom state influencing electrode.

Abstract: A computer-implemented method includes fetching, by a controller, data using a plurality of memory channels of a memory system. The method further includes detecting, by the controller, that a first memory channel of the plurality of memory channels has not returned data. The method further includes marking, by the controller, the first memory channel from the plurality of memory channels as unavailable. The method further includes, in response to a fetch, reconstructing, by the controller, fetch data based on data received from all memory channels other than the first memory channel.

Abstract: A method for testing a system under test (SUT) in an active environment includes receiving, by a testing system, a code path of the SUT that causes a soft failure in the active environment. The soft failure occurs in the active environment during execution of the SUT based at least on a parameter of the active environment. The method further includes generating, by the testing system, multiple tests for testing the SUT, the tests generated based on a coverage model of the SUT, wherein the coverage model uses several attributes. The method further includes selecting, by the testing system, from the generated tests, a set of tests that are associated with the code path. The method further includes executing, by the testing system, only the set of tests that are selected on the SUT to analyze a cause of the soft failure.

Abstract: Using a first trained residual neural network, an event analysis is generated from a set of event description encodings, an event description encoding in the set of event description encodings encoding a natural language description of an event, the event analysis comprising a first set of coordinates in an event space. Using a second trained residual neural network, the event analysis and a user description encoding, a personalization analysis is generated, the user description encoding a user interaction with the event, the personalization analysis comprising a component of a second set of coordinates in the event space. Using a combination of the event analysis and the personalization analysis, a first event component is activated, the first event component comprising a component of the event.

Abstract: A request may be identified having one or more constraints for accessing disaggregated resources in a computing environment. One or more resources in a plurality of disaggregated resources may be identified based on the request. Computing server instances may be dynamically orchestrated using the one or more resources in the plurality of disaggregated resources based on the one or more constraints.

Abstract: An example operation may include one or more of receiving, from a blockchain peer node, a sequence of blocks stored in a hash-linked chain of blocks on a distributed ledger, where each block in the sequence of blocks includes a reduced-step hash of block content from a previous block in the sequence, performing an approximate hash verification on the reduced-step hashes stored among the sequence of blocks, and determining whether the sequence of blocks has been tampered with based on the approximate hash verification on the reduced-step hashes.

Abstract: Machine learning model change management in an online Software as a Medical Device (“SaMD”) is provided. One or more machine learning models implemented in a medical domain may be monitored where the one or more machine learning models are continuously updated. One or more changes to the one or more machine learning models. The one or more machine learning models, having the one or more changes, are certified as being in compliance with performance characteristics and compliance criteria.

Abstract: A test apparatus is provided for use with a mainframe and an adapter. The test apparatus includes a logical adapter interface unit and a control system. The logical adapter interface unit is interposable between the adapter and the mainframe whereby an I/O signal transmittable from the adapter and to the mainframe is transmitted through the logical adapter interface unit. The logical adapter interface unit is configured to manipulate the I/O signal. The control system is coupled to the logical adapter interface unit and the mainframe and is configured to control manipulations of the I/O signal by the logical adapter interface unit to mimic a condition of I/O traffic being run through the adapter and to log a response of the mainframe to the manipulations.

Abstract: In one aspect of metadata track entry sorting in accordance with the present description, recovery logic sorts a list of metadata entries as a function of a source data track identification of each metadata entry to provide a second, sorted list of metadata entries, and generates a recovery volume which includes data tracks which are a function of one or more data target tracks identified by the sorted list of metadata entries. Because the metadata entry contents of the sorted list have been sorted as a function of source track identification number, the particular time version of a particular source track may be identified more quickly and more efficiently. As a result, recovery from data loss may be achieved more quickly and more efficiently thereby providing a significant improvement in computer technology. Other features and aspects may be realized, depending upon the particular application.