Abstract: An apparatus, system and method for protecting the confidentiality and integrity of a secure object running on a computer system by protecting the memory pages owned by the secure object, including assigning a secure object an ID, labeling the memory pages owned by a secure object with the ID of the secure object, maintaining an Access Control Monitor (ACM) table for the memory pages on the system, controlling access to memory pages by monitoring load and store instructions and comparing information in the ACM table with the ID of the software that is executing these instructions; and limiting access to a memory page to the owner of the memory page.

Abstract: Using a deep learning inference system, respective similarities are measured for each of a set of intermediate representations to input information used as an input to the deep learning inference system. The deep learning inference system includes multiple layers, each layer producing one or more associated intermediate representations. Selection is made of a subset of the set of intermediate representations that are most similar to the input information. Using the selected subset of intermediate representations, a partitioning point is determined in the multiple layers used to partition the multiple layers into two partitions defined so that information leakage for the two partitions will meet a privacy parameter when a first of the two partitions is prevented from leaking information. The partitioning point is output for use in partitioning the multiple layers of the deep learning inference system into the two partitions.

Abstract: Deep learning training service framework mechanisms are provided. The mechanisms receive encrypted training datasets for training a deep learning model, execute a FrontNet subnet model of the deep learning model in a trusted execution environment, and execute a BackNet subnet model of the deep learning model external to the trusted execution environment. The mechanisms decrypt, within the trusted execution environment, the encrypted training datasets and train the FrontNet subnet model and BackNet subnet model of the deep learning model based on the decrypted training datasets. The FrontNet subnet model is trained within the trusted execution environment and provides intermediate representations to the BackNet subnet model which is trained external to the trusted execution environment using the intermediate representations. The mechanisms release a trained deep learning model comprising a trained FrontNet subnet model and a trained BackNet subnet model, to the one or more client computing devices.

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 system and method for achieving power isolation across different cloud tenants and workloads is provided. The system includes a model of per-workload power consumption and an approach for attributing power consumption for each container. It allows a cloud provider to detect abnormally high power usage caused by specific containers and/or tenants, and to neutralize the emerging power attacks that exploit information leakages in the public container cloud. The approach also enables the provider to bill tenants for their specific power usage. Thus, the technique herein provides a mechanism that operates to attribute power consumption data for each container to defend against emerging power attacks, as well as to make it feasible to develop a cloud billing model based on power usage. The mechanism defends against emerging power attacks in container cloud offerings by implementing in a power-based namespace workflow in an OS kernel.

Abstract: A service processor is provided that includes a processor, a memory coupled to the processor and having instructions for executing an operating system kernel having an integrity management subsystem, secure boot firmware, and a tamper-resistant secure trusted dedicated microprocessor. The secure boot firmware performs a secure boot operation to boot the operating system kernel of the service processor. The secure boot firmware records first measurements of code executed by the secure boot firmware when performing the boot operation, in one or more registers of the tamper-resistant secure trusted dedicated microprocessor. The operating system kernel enables the integrity management subsystem. The integrity management subsystem records second measurements of software executed by the operating system kernel, in the one or more registers of the tamper-resistant secure trusted dedicated microprocessor.

Abstract: Handshake protocol layer features are extracted from training data associated with encrypted network traffic of a plurality of classified devices. Record protocol layer features are extracted from the training data. One or more models are trained based on the extracted handshake protocol layer features and the extracted record protocol layer features. The one or more models are applied to an observed encrypted network traffic stream associated with a device to determine a predicted device classification of the device.

Abstract: Mechanisms are provided to implement an enhanced privacy deep learning system framework (hereafter “framework”). The framework receives, from a client computing device, an encrypted first subnet model of a neural network, where the first subnet model is one partition of multiple partitions of the neural network. The framework loads the encrypted first subnet model into a trusted execution environment (TEE) of the framework, decrypts the first subnet model, within the TEE, and executes the first subnet model within the TEE. The framework receives encrypted input data from the client computing device, loads the encrypted input data into the TEE, decrypts the input data, and processes the input data in the TEE using the first subnet model executing within the TEE.

Abstract: Mechanisms are provided to implement an enhanced privacy deep learning system framework (hereafter “framework”). The framework receives, from a client computing device, an encrypted first subnet model of a neural network, where the first subnet model is one partition of multiple partitions of the neural network. The framework loads the encrypted first subnet model into a trusted execution environment (TEE) of the framework, decrypts the first subnet model, within the TEE, and executes the first subnet model within the TEE. The framework receives encrypted input data from the client computing device, loads the encrypted input data into the TEE, decrypts the input data, and processes the input data in the TEE using the first subnet model executing within the TEE.

Abstract: A method, system and apparatus for provisioning a computation into a trusted execution environment, including verifying the trusted execution environment, generating integrity information of the computation, generating sealed data, sending information of the computation, the sealed data, and integrity information to the trusted execution environment, confirming the sealed data, and verifying integrity of the computation information from the integrity information and the computation information.

Abstract: A method, system and apparatus for generating a computation such that it will execute in a target trusted execution environment (TEE), including selecting the target TEE, generating an authorization that is satisfied by a TEE, associating the authorization with the computation that executes in the TEE that is authorized, and generating the computation with the associated authorization.

Abstract: An example operation may include one or more one or more of receiving two or more authorization decisions from two or more authorization entities into a blockchain system, recording the two or more authorization decisions into one or more blocks of a blockchain of the blockchain system, determining, by the blockchain system, whether the two or more authorization decisions satisfy a policy to authorize access to at least one of a device or identifiable content on the device, and when the two or more authorization decisions satisfy the policy, authorizing access to a public key that can be used to gain access to the device.

Abstract: A method alters a computer resource while in a particular geolocation. One or more processors detect that a geolocation of a computer resource has changed from an initial geolocation to a new geolocation, where the computer resource is a virtual machine (VM). In response to detecting that the geolocation of the VM has changed from the initial geolocation to the new geolocation, the processor(s) reduce a functionality of the VM by reducing a quantity of processors and storage space available to the VM at the new geolocation as compared to the initial geolocation.

Abstract: Mechanisms for booting a service processor are provided. With these mechanisms, the service processor executes a secure boot operation of secure boot firmware to boot an operating system kernel of the service processor. The secure boot firmware records first measurements of code executed by the secure boot firmware when performing the boot operation, in one or more registers of a tamper-resistant secure trusted dedicated microprocessor of the service processor. The operating system kernel executing in the service processor enables an integrity management subsystem of the operating system kernel which records second measurements of software executed by the operating system kernel, in the one or more registers of the tamper-resistant secure trusted dedicated microprocessor.

Abstract: A method, apparatus, computer system, and computer program product for releasing secret information. A client on a computer system performing an attestation to a server on another computer system. The client receives an authorization that authorizes releasing the secret information. The client releases the secret information from a co-processor on the computer system using the authorization received from the server.

Abstract: A client seeking to establish a cryptographically-secure channel to a server has an associated public key acceptance policy. The policy specifies a required number of certificates that must be associated with the server's public key, as well as one or more conditions associated with those certificates, that must be met before the client “accepts” the server's public key. The one or more conditions typically comprise a trust function that must be satisfied before a threshold level of trust of the client is met. A representative public key acceptance policy would be that certificate chains for the public key are valid and non-overlapping with different root CAs, and that some configurable number of those chains be present. The technique may be implemented within the context of an existing client-server SSL/TLS handshake.

Abstract: A secure cloud computing environment protects the confidentiality of application code from a customer while simultaneously protecting the confidentiality of a customer's data from intentional or inadvertent leaks by the application code. This result is accomplished without the need to trust the application code and without requiring human surveillance or intervention. A client secure virtual machine (SVM) is accessible by a client who supplies commands, operand data and application data. An appliance SVM has the application code loaded therein and includes an application program interface that accesses a memory area shared by both SVMs. All access to the appliance SVM is initially revoked by an ultravisor, except for the shared memory and an encrypted persistent storage. The appliance SVM stores the application data in the persistent storage. The ultravisor manages an SVM by maintaining exclusive control over a device tree used by the operating system of the SVM.

Abstract: A method alters a computer resource in response to the computer resource moving from a first geolocation to a second geolocation. One or more processors receive a message indicating that a computer resource has moved from a first geolocation to a new geolocation. In response to receiving the message that the computer resource has moved from the first geolocation to the new geolocation, the processor(s) encrypt data that is stored on the computer resource, and apply decryption information to the encrypted data from the new geolocation, where the decryption information is specifically for decrypting encrypted data at the new geolocation. In response to the decryption information failing to decrypt the encrypted data at the new geolocation, the processor(s) and/or a user alter the computer resource.

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 processor-implemented method for a secure processing environment for protecting sensitive information is provided. The processor-implemented method may include receiving encrypted data and routing the encrypted data to the secure processing environment. Then the encrypted data may be decrypted and fields containing sensitive information may be found. The method may also include obfuscating the sensitive information and returning, by the secure processing environment, the decrypted data and obfuscated data.