Google Inc.

Mitsubishi Electric Research Laboratories, Inc.

Michael Andrew Terry

Andrei Kapishnikov

Tolga Bolukbasi

Fernanda Bertini Viégas

Kathleen Ryall

Carrie Cai

Tongshuang Wu

Methods, systems, devices, and tangible non-transitory computer readable media for saliency visualization are provided. The disclosed technology can include receiving a data input including a plurality of features. The data input can be segmented into regions. At least one of the regions can include two or more of the features. Attribution scores can be respectively generated for features of the data input. The attribution scores for each feature can be indicative of a respective saliency of such feature. A respective gain value for each region can be determined over one or more iterations based on the respective attribution scores associated with the features included in the region. Further, at each iteration one or more of the regions with the greatest gain values can be added to a saliency mask. Furthermore, at each iteration a saliency visualization can be produced based on the saliency mask.

Attribution and generation of saliency visualizations for machine-learning models

The present disclosure provides to transparent and controllable human-AI interaction via chaining of machine-learned language models. In particular, although existing language models (e.g., so-called 'large language models' (LLMs)) have demonstrated impressive potential on simple tasks, their breadth of scope, lack of transparency, and insufficient controllability can make them less effective when assisting humans on more complex tasks. In response, the present disclosure introduces the concept of chaining instantiations of machine-learned language models (e.g., LLMs) together, where the output of one instantiation becomes the input for the next, and so on, thus aggregating the gains per step.

Transparent and controllable human-AI interaction via chaining of machine-learned language models

A computerized method identifies entities accessing locations in a communications network. A unique identification is associated with a private identification for each entity accessing the network. A known list is maintained for each entity. Each entry in the known list for a particular entity includes the unique identification and private identification of other entities known to the particular entity. A location list is maintained. Each entry in the location list includes a plurality of unique identifications of other entities having shared a common location in the network. A message is generated only if at least two unique identifications of a particular entry of the location list are identical to at least two unique identifications found in the known list of the particular entity. The message includes the associated private identifications of the at least two identical unique identifications.

Network and method for identifying entities sharing a common network location

Growing community of inventors

Cambridge, MA, United States of America

Michael Andrew Terry