Abstract: Methods and systems are disclosed for managing access to encrypted data and encryption keys. The system stores, by a key management server, a first encryption key associated with a first service and a second encryption key associated with a second service. The system prevents, by the key management server, the second service from accessing the second encryption key while the first service is performing a first function using the first encryption key and determines that a first threshold period of time associated with the first function has elapsed. The system, in response to determining that the first threshold period of time associated with the first function has elapsed, prevents, by the key management server, the first service from accessing the first encryption key while the second service is performing a second function using the second encryption key.

Abstract: In one example an apparatus comprises a memory and a processor to create, from a first deep neural network (DNN) model, a first plurality of DNN models, generate a first set of adversarial examples that are misclassified by the first plurality of deep neural network (DNN) models, determine a first set of activation path differentials between the first plurality of adversarial examples, generate, from the first set of activation path differentials, at least one composite adversarial example which incorporates at least one intersecting critical path that is shared between at least two adversarial examples in the first set of adversarial examples, and use the at least one composite adversarial example to generate a set of inputs for a subsequent training iteration of the DNN model. Other examples may be described.

Abstract: The present disclosure is directed to systems and methods for the selective introduction of low-level pseudo-random noise into at least a portion of the weights used in a neural network model to increase the robustness of the neural network and provide a stochastic transformation defense against perturbation type attacks. Random number generation circuitry provides a plurality of pseudo-random values. Combiner circuitry combines the pseudo-random values with a defined number of least significant bits/digits in at least some of the weights used to provide a neural network model implemented by neural network circuitry. In some instances, selection circuitry selects pseudo-random values for combination with the network weights based on a defined pseudo-random value probability distribution.

Abstract: In some embodiments, an encryption system secures data using a homomorphic encryption. The encryption system encrypts a number by encrypting a number identifier of the number and combining the number and the encrypted number identifier using a mathematical operation to generate an encrypted number. The encrypted numbers may be stored at a server system along with their number identifiers. The server system can then generate an aggregation (e.g., sum) of the encrypted numbers and provide the aggregation, the encrypted numbers, and the number identifiers. The encryption system can then separate the aggregation of the numbers from the aggregation of the encrypted numbers using an inverse of the mathematical operation used in the encryption to effect removal of an aggregation of the encrypted number identifiers of the numbers from the aggregation of the encrypted numbers. The separated aggregation of the numbers is an aggregation of the plurality of the numbers.

Abstract: In one example an apparatus comprises a memory and a processor to create, from a first deep neural network (DNN) model, a first plurality of DNN models, generate a first set of adversarial examples that are misclassified by the first plurality of deep neural network (DNN) models, determine a first set of activation path differentials between the first plurality of adversarial examples, generate, from the first set of activation path differentials, at least one composite adversarial example which incorporates at least one intersecting critical path that is shared between at least two adversarial examples in the first set of adversarial examples, and use the at least one composite adversarial example to generate a set of inputs for a subsequent training iteration of the DNN model. Other examples may be described.

Abstract: The present disclosure is directed to systems and methods for the selective introduction of low-level pseudo-random noise into at least a portion of the weights used in a neural network model to increase the robustness of the neural network and provide a stochastic transformation defense against perturbation type attacks. Random number generation circuitry provides a plurality of pseudo-random values. Combiner circuitry combines the pseudo-random values with a defined number of least significant bits/digits in at least some of the weights used to provide a neural network model implemented by neural network circuitry. In some instances, selection circuitry selects pseudo-random values for combination with the network weights based on a defined pseudo-random value probability distribution.

Abstract: In some embodiments, an encryption system secures data using a homomorphic encryption. The encryption system encrypts a number by encrypting a number identifier of the number and combining the number and the encrypted number identifier using a mathematical operation to generate an encrypted number. The encrypted numbers may be stored at a server system along with their number identifiers. The server system can then generate an aggregation (e.g., sum) of the encrypted numbers and provide the aggregation, the encrypted numbers, and the number identifiers. The encryption system can then separate the aggregation of the numbers from the aggregation of the encrypted numbers using an inverse of the mathematical operation used in the encryption to effect removal of an aggregation of the encrypted number identifiers of the numbers from the aggregation of the encrypted numbers. The separated aggregation of the numbers is an aggregation of the plurality of the numbers.