International Business Machines Corporation

Marc Phillipe Stoecklin

Jialong Zhang

Teryl Paul Taylor

Zhongshu Gu

Frederico Araujo

Heqing Huang

A stackable filesystem architecture that curtails data theft and ensures file integrity protection. In this architecture, processes are grouped into ranked filesystem views, or 'security domains.' Preferably, an order theory algorithm is utilized to determine a proper domain in which an application is run. In particular, a root domain provides a single view of the filesystem enabling transparent filesystem operations. Each security domain transparently creates multiple levels of stacking to protect the base filesystem, and to monitor file accesses without incurring significant performance overhead. By combining its layered architecture with view separation via security domains, the filesystem maintains data integrity and confidentiality.

Filesystem view separation for data confidentiality and integrity using lattice-based security domains

A framework to accurately and quickly verify the ownership of remotely-deployed deep learning models is provided without affecting model accuracy for normal input data. The approach involves generating a watermark, embedding the watermark in a local deep neural network (DNN) model by learning, namely, by training the local DNN model to learn the watermark and a predefined label associated therewith, and later performing a black-box verification against a remote service that is suspected of executing the DNN model without permission. The predefined label is distinct from a true label for a data item in training data for the model that does not include the watermark. Black-box verification includes simply issuing a query that includes a data item with the watermark, and then determining whether the query returns the predefined label.

Growing community of inventors

White Plains, NY, United States of America

Marc Phillipe Stoecklin