Abstract: Methods and systems for managing artificial intelligence (AI) models are disclosed. To manage AI models, an instance of an AI model may not be re-trained using training data determined to be too similar to previously used training data. By doing so, malicious attacks intending to shift the AI model in a particular direction using poisoned training data may be prevented. To do so, a candidate training data set may be analyzed prior to performing re-training of an instance of an AI model using the candidate training data set. The analysis may result in a score. If the score exceeds a score threshold, the candidate training data set may be considered to contain poisoned training data. If the score does not exceed the score threshold, the candidate training data set may be accepted as usable to train an instance of the AI model.

Abstract: Methods and systems for securing deployments are disclosed. The deployments may be secured by generating and deploying security models to components of the deployment. The security models may be obtained through simulation of the operation of the deployment. During the simulation, different types of attacks on its operation and potential defenses to the attacks may be evaluated. The defenses able to defend against the different types of attacks may be used to generate the security models.

Abstract: Methods and systems for managing impact of inferences provided to inference consumers on decisions made by the inference consumers are disclosed. Poisoned training data may be introduced and used to train an AI model, which may then poison the AI model and lead to poisoned inferences being provided to the inference consumers. Inference consumers may deploy hardware to customers based on the poisoned inferences. To determine whether to modify the deployed hardware, a performance cost associated with the deployed hardware may be obtained. The performance cost may indicate a deviation between operation of the deployed hardware and operation of hardware that may have been deployed if an unpoisoned inference was used. If the performance cost meets a performance cost threshold, at least one additional hardware component may be deployed to the customer.

Abstract: Methods and systems for managing execution of inference models hosted by data processing systems are disclosed. To manage execution of inference models hosted by data processing systems, a system may include an inference model manager and any number of data processing systems. The inference model manager may communication system data for the communication system linking the data processing systems. The inference model manager may use the communication system data to determine whether the communication system meets inference generation requirements of the downstream consumer. If the communication system does not meet inference generation requirements of the downstream consumer, the inference model manager may obtain an inference generation plan to return to compliance with the inference generation requirements of the downstream consumer.

Abstract: Methods and systems for managing data used to provide computer-implemented services are disclosed. To manage the data, a request for corroborated data may be obtained from a data consumer, the request indicating a desired information content, corroboration requirements, and an urgency of need for the corroborated data. A determination may be made regarding whether a partial corroboration process is to be performed, the partial corroboration process being usable to obtain partially corroborated data. If the partial corroboration process is to be performed, the corroboration requirements may be relaxed to obtain relaxed corroboration requirements. The partial corroboration process may be performed to obtain partially corroborated data that meets the relaxed corroboration requirements and the urgency of need. At least a portion of the partially corroborated data may be provided to the data consumer to facilitate a partial provisioning of the computer-implemented services.

Abstract: Methods and systems for managing data used to provide computer-implemented services are disclosed. To manage the data, a request for corroborated data may be obtained from a data consumer indicating a desired information content and corroboration requirements for the corroborated data. Based on the request, the corroborated data may be obtained which has the desired information content and may be obtained by a first data source. The corroborated data may be corroborated using at least second data obtained by a second data source adapted to measure a similar information content to the desired information content. The corroborated data may be corroborated using relaxed corroboration requirements due to at least a portion of the corroboration requirements being unable to be met. At least a portion of the corroborated data may be provided to the data consumer to facilitate provisioning of the computer-implemented services.

Abstract: Methods and systems for managing data used to provide computer-implemented services are disclosed. To manage the data, a request for corroborated data may be obtained from a data consumer indicating a desired information content and a threshold level of trust for the corroborated data. Based on the request, the corroborated data may be obtained from a corroborated data database. The corroborated data may have the desired information content and may be ascribed a level of trust that meets the level of trust threshold. The corroborated data may also have been obtained by a first data source, corroborated using at least a second data source, and corroborated based on a corroboration plan and a dynamically updated list of data sources. At least a portion of the corroborated data may be provided to the data consumer to facilitate provisioning of the computer-implemented services.

Abstract: Methods and systems for verifying data used to provide computer-implemented services as non-synthetic data without obtaining a copy of the data are disclosed. To do so, a data poisoning pattern may be provided for use in obtaining poisoned data using the data and a hash of the poisoned data may be obtained. In response to obtaining the hash of the poisoned data, an inference generation process may be initiated to obtain an inference generated by an inference model and a second hash of the poisoned data generated by the inference model. The inference model may be trained to identify data poisoning patterns using poisoned data as ingest. If the second hash matches the hash and the inference correctly identifies the data poisoning pattern, it may be determined that the data is verified as non-synthetic data. The hash of the poisoned data may then be stored in a data repository.

Abstract: Methods and systems for managing data used to provide computer-implemented services are disclosed. To manage the data, a first request for corroborated data may be obtained from a data consumer indicating a desired information content and a threshold level of trust. Based on the request, the corroborated data may be obtained which has the desired information content and a level of trust that meets the level of trust threshold. The corroborated data may be obtained by a first data source and may be corroborated using at least second data obtained by a second data source in a sandboxed environment that maintains confidentiality of the second data. The second data source may be adapted to measure a similar information content to the desired information content. At least a portion of the corroborated data may be provided to the data consumer to facilitate provisioning of the computer-implemented services.

Abstract: Methods and systems for managing inference models hosted by data processing systems are disclosed. To manage the inference models, data processing systems that host the inference models may be monitored for risk of compromise. The data processing systems may be monitored by evaluating the processing results generated by the data processing systems. When a data processing system is identified as being compromised, a self-healing process or other remediation process may be initiated to reduce or eliminate the threat to successful inference generation presented by the compromised data processing system.

Abstract: Methods, systems, and devices for providing computer-implemented services are disclosed. To provide the computer-implemented services, inference models may be deployed to locations. Prior to deploying an inference model to a location, it may be determined whether the location is trustworthy. If the location is determined to not be trustworthy, an input data attack resistant inference model may be selected and deployed. The input data attack resistant inference model may be trained to impart reconstruction resistance to sensitive information during inference generation based on a schema for weighting the sensitive information for reconstruction resistance. The training process may decrease a likelihood of the inferences generated by the input data attack resistant inference model being usable to reconstruct the sensitive information.

Abstract: Methods and systems for managing an attack on an artificial intelligence (AI) model using view level analysis are disclosed. As AI models are updated over time using new training data, snapshots of the AI models may be obtained. The snapshots may include information regarding the training data used to train the AI model, the parameters of the AI model, and/or the inferences obtained from the AI model. A malicious party may perform an attack on the AI model by introducing poisoned training data through a data source. The content of supplied poisoned training data may be determined based on a view level into the AI model. The view level of the malicious party may be used to design countermeasures to mitigate and/or prevent future attacks to the AI model.

Abstract: Methods and systems for managing data used to provide computer-implemented services are disclosed. To manage the data, a request for corroborated data may be obtained from a data consumer indicating a desired information content and a threshold level of trust for the corroborated data. Based on the request, the corroborated data may be obtained from a corroborated data database. The corroborated data may be based on at least a pre-computed partial corroboration result and a new partial corroboration result having been computed dynamically at a time of corroboration of the corroborated data. At least a portion of the corroborated data may be provided to the data consumer to facilitate provisioning of the computer-implemented services.

Abstract: Methods and systems for managing data used to provide computer-implemented services are disclosed. To manage the data, a request for corroborated data may be obtained from a data consumer indicating a desired information content and a threshold level of trust. Based on the request, the corroborated data may be obtained from a corroborated data database which has the desired information content and a level of trust that meets the level of trust threshold. The corroborated data may be obtained from a first data source and may be corroborated using at least one other data source adapted to measure a similar information content to the desired information content which the first data source may be adapted to measure. The level of trust may be based on a corroboration schema. At least a portion of the corroborated data may be provided to the data consumer to facilitate provisioning of the computer-implemented services.

Abstract: Methods and systems for managing methods of communication used by data processing systems throughout a distributed environment are disclosed. To manage methods of communication, a system may include a communication manager and one or more data processing systems. The communication manager may monitor use of an existing method of communication by a data processing system over time. If characteristics of the existing method of communication do not meet criteria specified by a communication requirement, the communication manager may determine whether to update the existing method of communication. To do so, the communication manager may host and operate a digital twin of the data processing system and may simulate operation of the data processing system using other methods of communication using the digital twin. If one of the other methods of communication has characteristics that meet the communication requirement, the communication manager may update the existing method of communication.

Abstract: Methods and systems for securing deployments are disclosed. The deployments may be secured by generating and deploying security models to components of the deployment. The security models may be obtained through simulation of the operation of the deployment. During the simulation, predictions of different types of attacks and the potential defenses to the attacks on its operation may be evaluated. Further, limits may be imposed on the different attacks and potential defenses to simulate various scenarios that may be encountered in real systems.

Abstract: Methods and systems for identifying a goal of a malicious party that poisoned an artificial intelligence (AI) model are disclosed. As AI models are updated over time using new training data, snapshots of the AI models may be obtained. The snapshots may include information regarding the structure of the AI model, the inferences obtained from the AI model (e.g., outcomes of the inferences on inference consumers), and/or the training data used to train the AI model instance (e.g., data sources that supply the training data). A malicious party may introduce poisoned training data to an AI model to control the outcomes of consumed inferences on inference consumers. To identify the goal of the malicious party and/or to identify the malicious party itself, the relationships of information from the snapshots and the outcomes for inference consumer groups may be analyzed.

Abstract: Methods, systems, and devices for providing computer-implemented services are disclosed. To provide the computer-implemented services, inference models may be deployed to locations. Prior to deploying an inference model to a location, it may be determined whether the location is trustworthy. If the location is determined to not be trustworthy, an input data attack resistant inference model may be selected and deployed. The input data attack resistant inference model may be trained to impart reconstruction resistance to sensitive information during inference generation based on a schema for weighting the sensitive information for reconstruction resistance. The training process may decrease a likelihood of the inferences generated by the input data attack resistant inference model being usable to reconstruct the sensitive information.

Abstract: Methods and systems for managing a generative inference model are disclosed. Using the generative inference model and ingest data, a graphical inference may be obtained. A structured representation for the graphical inference may be populated based on objects and/or stylistic elements displayed by the graphical inference, and may indicate (e.g., when compared to structured representations for the reference works) whether the graphical inference exceeds a predetermined level of similarity with respect to the reference works. If the inference exceeds the predetermined level of similarity, performance of an action set may be initiated to manage an impact of similarities between the graphical inference and the reference works.

Abstract: Methods, systems, and devices for providing computer-implemented services are disclosed. To provide the computer-implemented services, inference models may be deployed to locations to generate inferences. Prior to deploying an inference model to a location, it may be determined whether the location is trustworthy. If the location is determined to not be trustworthy, an input data attack resistant inference model may be selected and deployed. The input data attack resistant inference model may be based, at least in part, on a training process that includes a weight freezing process based on levels of reconstructability of input features based on inferences generated by the input data attack resistant inference model. The training process may decrease a likelihood of the inferences generated by the input data attack resistant inference model being usable to reconstruct input data used to generate the inferences. By doing so, sensitive input data may be protected during inference generation.