Aol Inc.

Nvidia Corporation

Citrix Systems, Inc.

Bright Sun Technologies

Internal Revenue Service United States Department of the Treasury

Yan Cheng

Zhihong Zhang

Andrew Feng

Daguang Xu

Philip W Flack

Matthew Nguyen

Fausto Milletari

Wenqi Li

Rong Ou

Nicola Christin Rieke

Larry L Lu

Jian Wang

Wentao Zhu

Zhenqiang Yu

Maximilian Baust

Charles Jonathan Hancox

Apparatuses, systems, and techniques to perform federated training of neural networks while maintaining control over dissemination of local models of neural networks from which aspects of local training data might be extracted. In at least one embodiment, a neural network is trained on local training data and a local model is provided to be aggregated with other local models into a global model that is in turn used for further local model training, wherein a provided local model or training is adjusted to reduce an ability to extract aspects of local training data therefrom.

Neural network trained using federated learning with local training data preserved at local edge circuits

Apparatuses, systems, and techniques to collaboratively train one or more machine learning models. Parameter reviewers may be configured to compare sets of machine learning model parameter information in order to generate one or more machine learning models, such as neural networks.

Processor and system to train machine learning models based on comparing accuracy of model parameters

The present invention is a multi-layer computer architecture which separately extracts ALC business and logical functions and data. The architecture creates a Java object model or other target language object model which allows comparison of ALC data with target language data to verify logical processes. These object models can be directly traced back to the legacy ALC. The data model is automatically generated from a scan of the ALC and leverages generic patterns which can be reused to generate Java representations of other legacy code bases.

Layered software architecture model for translation of assembler language to target language

Techniques for annotating communications are described. An electronic message sent from a sender messaging identity to a recipient messaging identity is received. The message is analyzed to identify third-party information related to a portion of the message. The portion of the message is annotated to enable presentation of the third-party information by the recipient messaging identity. The message including the annotated portion of the message is presented on a display device for viewing by the recipient messaging identity.

Annotated communication

Techniques are provided for improving security in a single-sign-on context by providing, to a user's client system, two linked authentication credentials in separate logical communication sessions and requiring that both credentials be presented to a host system. Only after presentation of both credentials is the user authenticated and permitted to access applications on the host system.

Authenticating a client using linked authentication credentials

Techniques are described for repairing some types of user account problems that interfere with granting a user access to a computer system and doing so during a process to authenticate the user in a way that does not require the user to re-enter authentication information or require the user to restart a communication session with the computer system. In response to a determination that a user's account has a problem during an authentication process, techniques are provided to enable a user to execute an appropriate process or processes to fix the user account, after which the authentication process continues. In this way, the correction to the user account may appear to be seamless to the user.

Modifying a user account during an authentication process

Growing community of inventors

Great Falls, VA, United States of America

Yan Cheng