Abstract: In one embodiment, a device may identify one or more vulnerable portions of a program to be observed based on security vulnerability information. The device may instrument the program with an observability control to configure collecting of observability information regarding the one or more vulnerable portions of the program. The device may modify the observability control based on one or more attributes associated with the collecting of the observability information regarding the one or more vulnerable portions of the program. The device may the observability information regarding the one or more vulnerable portions of the program according to the observability control as modified.

Abstract: A deep-learning based method evaluates similarities of entities in decentralized identity graphs. One or more processors represent a first identity profile as a first identity graph and a second identity profile as a second identity graph. The processor(s) compare the first identity graph to the second identity graph, which are decentralized identity graphs from different identity networks, in order to determine a similarity score between the first identity profile and the second identity profile. The processor(s) then implement a security action based on the similarity score.

Abstract: A computer-implemented method, system, and/or computer program product controls a driving mode of a self-driving vehicle (SDV). One or more processors compare a control processor competence level of an on-board SDV control processor in controlling the SDV to a human driver competence level of a human driver in controlling the SDV while the SDV encounters a current roadway condition which is a result of current weather conditions of the roadway on which the SDV is currently traveling. One or more processors then selectively assign control of the SDV to the SDV control processor or to the human driver while the SDV encounters the current roadway condition based on which of the control processor competence level and the human driver competence level is relatively higher to one another.

Abstract: A computer-implemented method in accordance with one embodiment includes, in response to a submission of an input dataset to an artificially intelligent application, receiving an explanation from each module of the application. The modules are configured within the application in a serial sequence in which each module, upon receiving the input dataset and any input generated by an immediately preceding module of the serial sequence, generates output that is forwarded as input to a next module, if any, in the sequence. A determination is made that at least two of the received explanations are semantically inconsistent.

Abstract: Anomalous control and data flow paths in a program are determined by machine learning the program's normal control flow paths and data flow paths. A subset of those paths also may be determined to involve sensitive data and/or computation. Learning involves collecting events as the program executes, and associating those event with metadata related to the flows. This information is used to train the system about normal paths versus anomalous paths, and sensitive paths versus non-sensitive paths. Training leads to development of a baseline “provenance” graph, which is evaluated to determine “sensitive” control or data flows in the “normal” operation. This process is enhanced by analyzing log data collected during runtime execution of the program against a policy to assign confidence values to the control and data flows. Using these confidence values, anomalous edges and/or paths with respect to the policy are identified to generate a “program execution” provenance graph associated with the policy.

Abstract: A computer-implemented method, system, and/or computer program product controls a driving mode of a self-driving vehicle (SDV). One or more processors compare a control processor competence level of an on-board SDV control processor in controlling the SDV to a human driver competence level of a human driver in controlling the SDV while the SDV encounters a current roadway condition which is a result of current weather conditions of the roadway on which the SDV is currently traveling. One or more processors then selectively assign control of the SDV to the SDV control processor or to the human driver while the SDV encounters the current roadway condition based on which of the control processor competence level and the human driver competence level is relatively higher to one another.

Abstract: A method, computer system, and a computer program product for text bias identification and correction are provided. A first text corpus may be received. A designation of a second text corpus may be received. Words of the first text corpus may be embedded as a first word embedding in an embedding model. The first word embedding may be compared to a second word embedding in the embedding model to identify a first biased text in the first text corpus. The second word embedding may be from the second text corpus. A first replacement text portion may be generated as a substitute for the first biased text. The first replacement text portion may include a first unbiased text. The first biased text and the first replacement text portion may be presented.

Abstract: In one embodiment, a device identifies a plurality of nodes of a distributed or federated learning system. The device receives model training results from the plurality of nodes. The device determines, based in part on the model training results or information about the plurality of nodes, whether a particular node or subset of nodes in the plurality of nodes provided fraudulent model training results. The device initiates a corrective measure with respect to the particular node or subset of nodes, based on a determination that the particular node or subset of nodes provided fraudulent model training results, in accordance with a policy.

Abstract: Preserving distributions of data values of a data asset in a data anonymization operation is provided. Anonymizing data values is performed by transforming sensitive data in a set of columns over rows of the data asset while preserving distribution of the data values in the set of transformed columns to a defined degree using a set of autoencoders and loss function. The autoencoders are base trained from preexisting data in a data assets catalog and actively trained during data dissemination. Parametric coefficients of the loss function are configured and the threshold is generated using policies from an enforcement decision for the data asset and data consumer. The loss function value of a selected row is compared to the threshold. Transformed data values of the selected row are transcribed to an output row when the loss function value is greater than the threshold and disseminated to the data consumer.

Abstract: Smart speaker system mechanisms, associated with a smart speaker device comprising an audio capture device, are provided for processing audio sample data captured by the audio capture device. The mechanisms receive, from the audio capture device of the smart speaker device, an audio sample captured from a monitored environment. The mechanisms classify a sound in the audio sample data as a type of sound based on performing a joint analysis of a plurality of different characteristics of the sound and matching results of the joint analysis to criteria specified in a plurality of sound models. The mechanisms determine, based on the classification of the sound, whether a responsive action is to be performed based on the classification of the sound. In response to determining that a responsive action is to be performed, the mechanisms initiate performance of the responsive action by the smart speaker system.

Abstract: A computer-implemented method, system, and/or computer program product controls a driving mode of a self-driving vehicle (SDV). One or more processors compare a control processor competence level of an on-board SDV control processor in controlling the SDV to a human driver competence level of a human driver in controlling the SDV while the SDV encounters a current roadway condition which is a result of current weather conditions of the roadway on which the SDV is currently traveling. One or more processors then selectively assign control of the SDV to the SDV control processor or to the human driver while the SDV encounters the current roadway condition based on which of the control processor competence level and the human driver competence level is relatively higher to one another.

Abstract: A method provides a security action based on identity profile scores. One or more processors represent an identity profile as a knowledge graph. The processor(s) associate a set of changes of the identity profile across a plurality of identity networks with a fraud score. The processor(s) then implement a security action based on the fraud score.

Abstract: An example operation may include one or more of creating, by a blockchain user of a blockchain network, a world state checkpoint transaction requesting world state validation, endorsing, by one or more endorser nodes or peers, the world state checkpoint transaction, transferring endorsements to the blockchain user, recording, by an orderer node or peer, the endorsed world state checkpoint transaction into a block, validating and committing all transactions in the block, calculating and signing a hash of a current world state, by all blockchain nodes or peers of the blockchain network, and verifying, by the blockchain user, world state integrity from the calculated and signed hashes of the current world state.

Abstract: A secure chain of data blocks is maintained at a given computing node. The given computing node is part of a set of computing nodes in a distributed network of computing nodes wherein each of the set of computing nodes maintains the secure chain of data blocks. The secure chain of data blocks maintained at each computing node comprises one or more data blocks that represent one or more accident related transactions associated with a vehicle. In response to a risk assessment operation, one or more data blocks are added to the secure chain of data blocks maintained at the given computing node.

Abstract: A computer-implemented method in accordance with one embodiment includes, in response to a submission of an input dataset to an artificially intelligent application, receiving an explanation from each module of the application. The modules are configured within the application in a serial sequence in which each module, upon receiving the input dataset and any input generated by an immediately preceding module of the serial sequence, generates output that is forwarded as input to a next module, if any, in the sequence. A determination is made that at least two of the received explanations are semantically inconsistent.

Abstract: A computer-implemented method, system, and/or computer program product controls self-driving vehicles (SDVs). An emergency message is transmitted to a receiver within a self-driving vehicle (SDV). The emergency message describes an emergency state of an emergency vehicle and an identified future route of the emergency vehicle, where the identified future route is a planned route to an emergency destination for the emergency vehicle that includes a first pathway. In response to the SDV receiving the emergency message, the SDV is redirected, via an auto-control hardware system on the SDV, to drive to a second pathway that does not conflict with the identified future route of the emergency vehicle.

Abstract: A computer implemented method and related apparatus defend a system against adversarial queries. An enforcement graph is provided and used to enforce data policies for a system. A generative adversarial model (GAN) is used for querying the enforcement graph to detect a potential adversarial query-based attack against the enforcement graph A policy is provided to protect the enforcement graph against the potential adversarial attack.

Abstract: Preserving distributions of data values of a data asset in a data anonymization operation is provided. Anonymizing data values is performed by transforming sensitive data in a set of columns over rows of the data asset while preserving distribution of the data values in the set of transformed columns to a defined degree using a set of autoencoders and loss function. The autoencoders are base trained from preexisting data in a data assets catalog and actively trained during data dissemination. Parametric coefficients of the loss function are configured and the threshold is generated using policies from an enforcement decision for the data asset and data consumer. The loss function value of a selected row is compared to the threshold. Transformed data values of the selected row are transcribed to an output row when the loss function value is greater than the threshold and disseminated to the data consumer.

Abstract: An apparatus and related method defend against adversarial queries. A policy enforcement hypergraph is constructed to express a set of security policies. Then, the hypergraph is repeatedly traversed to determine whether a user behavior is changing over time. The user behavior is measured by reference to a vertex or an edge in the hypergraph. If it is determined that the user behavior has changed over time an enforcement action is taken based on a security policy.

Abstract: An explainable artificially intelligent (XAI) application contains an ordered sequence of artificially intelligent software modules. When an input dataset is submitted to the application, each module generates an output dataset and an explanation that represents, as a set of Boolean expressions, reasoning by which each output element was chosen. If any pair of explanations are determined to be semantically inconsistent, and if this determination is confirmed by further determining that an apparent inconsistency was not a correct response to an unexpected characteristic of the input dataset, nonzero inconsistency scores are assigned to inconsistent elements of the pair of explanations.