Abstract: Privacy-preserving reconstruction for compressed sensing is described. An example of a method includes capturing raw image data for a scene with a compressed sensing image sensor; performing reconstruction of the raw image data, including performing an enhancement reconstruction of the raw image data; and generating a masked image from the reconstruction of the raw image data, wherein the enhancement reconstruction includes applying enhancement utilizing a neural network trained with examples including image data in which private content is masked.

Abstract: An apparatus is disclosed. The apparatus comprises one or more processors to receive trained model update data from each of a plurality of collaborators, execute an auxiliary machine learning model to the trained model update data to generate a risk score for trained model update data associated with each collaborator, apply one or more policies based on the risk scores to generate adjusted trained model update data associated with each collaborator.

Abstract: Machine learning fraud resiliency using perceptual descriptors is described. An example of a computer-readable storage medium includes instructions for accessing multiple examples in a training dataset for a classifier system; calculating one or more perceptual hashes for each of the examples; generating clusters of perceptual hashes for the multiple examples based on the calculation of the one or more perceptual hashes for each of the plurality of examples; obtaining an inference sample for classification by the classifier system; generating a first classification result for the inference sample utilizing a neural network classifier and generating a second classification result utilizing the generated clusters of perceptual hashes; comparing the first classification result with the second classification result; and, upon a determination that the first classification result does not match the second classification result, determining a suspicion of an adversarial attack.

Abstract: An apparatus is disclosed. The apparatus comprises one or more processors to generate measurement matrices for a plurality of edge devices in a federated learning system, transmit the matrices to the plurality of edge devices, receive sampled trained model update data from the plurality of edge devices and reconstruct the sampled trained model update data using the measurement matrices to generate the trained model update data.

Abstract: Methods and apparatus for hardware based file/document expiry timer enforcement is disclosed. An example method includes instructing, by executing an instruction with a processor, a trusted execution environment to generate an encryption key and a certificate for a document, the certificate including expiry information for the document, the certificate associated with identification information of the document, and the expiry information indicative of a time period for which the encryption key is valid to decrypt the document; encrypting, by executing an instruction with the processor, the document using the encryption key; transmitting the certificate to a first remote network storage device; and transmitting the document to a second remote network storage device.

Abstract: In one example an apparatus comprises a memory and a processor to receive, in an edge node of a secure network, a first file, determine that the first file is addressed to a recipient outside the secure network, and in response to a determination that the first file is addressed to a destination outside the secure network, to generate a watermark that identifies a transmitter of the document, a recipient of the document, and comprises a digital signature of the first file, embed the watermark in the first file to generate a watermarked file, and pass the watermarked file to an input/output system for transmission out of the secure network. Other examples may be described.

Abstract: Technologies to facilitate supervision of an online identify include a gateway server to facilitate and monitor access to an online service by a user of a “child” client computer device. The gateway server may include an identity manager to receive a request for access to the online service from the client computing device, retrieve access information to the online service, and facilitate access to the online service for the client computing device using the access information. The access information is kept confidential from the user. The gateway server may also include an activity monitor module to control activity between the client computing device and the online service based on the set of policy rules of a policy database. The gateway server may transmit notifications of such activity to a “parental” client computing device for review and/or approval, which also may be used to update the policy database.

Abstract: Embodiments are directed to security optimizing compute distribution in a hybrid deep learning environment. An embodiment of an apparatus includes one or more processors to determine security capabilities and compute capabilities of a client machine requesting to use a machine learning (ML) model hosted by the apparatus; determine, based on the security capabilities and based on exposure criteria of the ML model, that one or more layers of the ML model can be offloaded to the client machine for processing; define, based on the compute capabilities of the client machine, a split level of the one or more layers of the ML model for partition of the ML model, the partition comprising offload layers of the one or more layers of the ML model to be processed at the client machine; and cause the offload layers of the ML model to be downloaded to the client machine.

Abstract: A ground transporter transports a package to a package recipient location and receives a token from a token recipient, wherein receiving the token represents a verification that the ground transporter has moved to the package recipient location. The ground transporter is configured, if the token is received, to deliver the package at the package recipient location.

Abstract: In one example an apparatus comprises a memory and a processor to receive, in an edge node of a secure network, a first file, determine that the first file is addressed to a recipient outside the secure network, and in response to a determination that the first file is addressed to a destination outside the secure network, to generate a watermark that identifies a transmitter of the document, a recipient of the document, and comprises a digital signature of the first file, embed the watermark in the first file to generate a watermarked file, and pass the watermarked file to an input/output system for transmission out of the secure network. Other examples may be described.

Abstract: An example apparatus includes: memory; instructions in the apparatus; and at least one processor to execute the instructions to: check for proof of trust information in one or more pre-determined positions in a trusted digital image, the proof of trust information including a secure output marker, the secure output marker indicative of information corresponding to a trusted output area of the trusted digital image; decrypt the secure output marker using one or more security keys from a trusted execution environment (TEE), the TEE isolated from a computing application; and enable activation of a trusted output indicator in response to a match between first data corresponding to the secure output marker and second data corresponding to the trusted output area of the trusted digital image.

Abstract: Techniques and apparatuses to harden AI systems against various attacks are provided. Among the different techniques and apparatuses, is provided, techniques and apparatuses that expand the domain for an inference model to include both visible classes and well as hidden classes. The hidden classes can be used to detect possible probing attacks against the model.

Abstract: Methods and apparatus for hardware based file/document expiry timer enforcement is disclosed. An example method includes instructing, by executing an instruction with a processor, a trusted execution environment to generate an encryption key and a certificate for a document, the certificate including expiry information for the document, the certificate associated with identification information of the document, and the expiry information indicative of a time period for which the encryption key is valid to decrypt the document; encrypting, by executing an instruction with the processor, the document using the encryption key; transmitting the certificate to a first remote network storage device; and transmitting the document to a second remote network storage device.

Abstract: The disclosed embodiments generally relate to detecting malware through detection of micro-architectural changes (morphing events) when executing a code at a hardware level (e.g., CPU). An exemplary embodiment relates to a computer system having: a memory circuitry comprising an executable code; a central processing unit (CPU) in communication with the memory circuitry and configured to execute the code; a performance monitoring unit (PMU) associated with the CPU, the PMU configured to detect and count one or more morphing events associated with execution of the code and to determine if the counted number of morphine events exceed a threshold value; and a co-processor configured to initiate a memory scan of the memory circuitry to identify a malware in the code.

Abstract: Partitioning a deep neural network (DNN) model into one or more sets of one or more private layers and one or more sets of one or more public layers, a set of one or more private layers being at least one key in a cryptographic system; and deploying the partitioned DNN model on one or more computing systems.

Abstract: Providing optical watermark signals for a visual authentication session by performing at least the following: receive, at an anti-spoof engine, an instruction to perform visual authentication operations for a visual authentication session, generate, with the anti-spoof engine, an optical watermark signal based on receiving the instruction, wherein the optical watermark signal includes at least one optical identifier to authenticate images captured during the visual authentication session, obtain, with the anti-spoof engine, an image source that includes captured images of the visual authentication session, determine, with the anti-spoof engine, whether the image source includes a reflected optical watermark signal, and compare, with the anti-spoof engine, whether the reflected optical watermark signal matches the generated optical watermark signal based on the determination that the image source includes the reflected optical watermark signal.

Abstract: Technologies to facilitate supervision of an online identify include a gateway server to facilitate and monitor access to an online service by a user of a “child” client computer device. The gateway server may include an identity manager to receive a request for access to the online service from the client computing device, retrieve access information to the online service, and facilitate access to the online service for the client computing device using the access information. The access information is kept confidential from the user. The gateway server may also include an activity monitor module to control activity between the client computing device and the online service based on the set of policy rules of a policy database. The gateway server may transmit notifications of such activity to a “parental” client computing device for review and/or approval, which also may be used to update the policy database.

Abstract: According to various examples, an automated delivery device is described including a transportation system configured to move a package to a package recipient location, a package drop-off mechanism configured to drop off the package at the package recipient location, a communication system configured to communicate with a recipient camera device at the package recipient location and a controller configured to instruct the recipient camera device via the transceiver to verify package delivery based on an image of the delivered package.

Abstract: An example display device for securing graphics outputs includes: a checker to check whether a secure output marker is located in a pre-determined position in graphical information of a digital image; a verifier to verify whether first data in the secure output marker matches actual data in a trusted output area of the digital image; and autonomous indicator control logic to activate a hardware-based trusted output indicator when the first data matches the actual data, the autonomous indicator control logic not accessible by computing applications executing on a system in communication with the display device.

Abstract: A system and method of detecting and remediating attacks includes receiving operating system (OS) read/write data from an OS, the OS read/write data describing at least one of reads from and writes to a storage device over a file system interface of the OS; collecting storage device read/write data, the storage device read/write data describing at least one of reads from and writes to the storage device; comparing the OS read/write data to the storage device read/write data; and determining if there is a discrepancy between the OS read/write data and the storage device read/write data. If there is a discrepancy, determining if there is an anomaly detected between OS read/write data and the storage device read/write data. If there is an anomaly, causing a remediation action to be taken to stop a malware attack.