Amazon Technologies, Inc.

Rahul Gupta

Martin Peter Aalund

Sachin Purushottam Joglekar

Arivazhagan Chandrashekaran

Melanie C B Gens

Hung-Bing Tan

Kristina Perez De Tagle

Sudarshan Rangaraj

Pierre Della Nave

Oszkar Tiberius Bajko

Mona Mayeh

Jimit Majmudar

Clement Chung

Christophe Dupuy

Ben Smith

Spyridon Matsoukas

Premkumar Natarajan

Anil Ramakrishna

Anoop Kumar

Xinhong Zhang

Sriram Venkatapathy

Pavel Bhowmik

Pooja Sanjay Sonawane

Samuel Harry Ingbar

Srinivasan Rengarajan Veeravanallur

Seyed Omid Razavi

Peyman Passban

Gazelle Tanya Roostaeyan

Lili Gehorsam

Sachin Midha

Jacob Ryan Stolee

Yuval Merhav

Zefei Li

Heather Brooke Spetalnick

Mayur Himmatbhai Dabhi

Qingwen Ye

Apurv Verma

Jwala Dhamala

Avni Khatri

A device includes a housing having a top, a bottom opposite the top, a front, and a back opposite the front. One or more microphones are oriented substantially towards the top. A first camera, a display, a loudspeaker, and an emitter are oriented substantially towards the front. A second camera, a projector, and a sensor are oriented substantially towards the bottom. The second camera and the sensor may include field of views that at least partially overlap with a projection area of the projector.

Audiovisual device

Devices and techniques are generally described for security threat mitigation for generative machine learning models. In some examples, first request data including a first request may be received. First prompt data may be generated based at least in part on the first request data. First plan data may be generated, the first plan data including a first API call to a first API of a first computer-implemented service. The first API call may be executed and first result data may be received in response to the first API call. A determination may be made that the first result data includes a first impermissible instruction to inject data into a subsequent prompt. First output data may be generated indicating that the first request cannot be completed.

Security threat mitigation for large language models

A system may share updates to a machine learning model between a first device and a second device. The first device may update parameters of a model operating on the first device in response to processing input data. The first device may generate model update data based on the updated parameters, and send the model update data to the second device. The first and second devices may agree on one or more protocols for generating the model update data to obscure details of the model, and/or reduce a communication cost of sending the model update data (e.g., relative to a size of the model). The second device may aggregate model update data (e.g., from multiple devices), determine second model update data, and send the second model update data to the first device. The first device may update its own model based on the second model update data.

Updating machine learning models

Devices and techniques are generally described for security threat mitigation for generative machine learning models. In some examples, first prompt data including first data associated with a first natural language input and a first span may be determined. An LLM may determine first plan data using the first prompt data. The first plan data may include a call to the first API. A first classifier model may determine a first trust score for the first span. A first attention score may be determined for the first span and the first action plan. Second plan data may be generated based on at least one of the first trust score and the first attention score or the second trust score and the second attention score.

Security for generative models using attention analysis

Techniques for evaluating a natural language understanding (NLU) component and determining an action to resolve an issue processing a user input are described. The system determines which component is invoked by a baseline NLU component is processing the user input, and which component is invoked by an updated NLU component. Based on that information, the system selects the action to resolve the updated NLU component generating an undesired response to the user input.

Evaluating natural language processing components

Devices and techniques are generally described for cohort determination in natural language processing. In various examples, a first natural language input to a natural language processing system may be determined. The first natural language input may be associated with a first account identifier. A first machine learning model may determine first data representing one or more words of the first natural language input. A second machine learning model may determine second data representing one or more acoustic characteristics of the first natural language input. Third data may be determined, the third data including a predicted performance for processing the first natural language input by the natural language processing system. The third data may be determined based on the first data representation and the second data representation.

Cohort determination in natural language processing

Techniques for updating a machine learning (ML) model are described. A device or system may receive input data corresponding to a natural or non-natural language (e.g., gesture) input. Using a first ML model, the device or system may determine the input data corresponds to a data category of a plurality of data categories. Based on the data category, the device or system may select a ML training type from among a plurality of ML training types. Using the input data, the device or system may perform the selected ML training type with respect to a runtime ML model to generate an updated ML model.

Machine learning model updating

A device include a first camera, a second camera, and a shutter assembly. The shutter assembly includes a switch, a link coupled to the switch and having a first cover, and an arm coupled to the link and having a second cover. Actuating the switch to a first position disposes the first cover within a first field of view of the first camera and the second cover within a second field of view of the second camera. Actuating the switch to a second position disposes the first cover outside of the first field of view and the second cover outside of the second field of view.

Electronic device with shutter assembly

Techniques for partially processing an input on a device and completing processing at a remote system are provided. The device may process an input using an on-device machine learning (ML) model, and determine to cease processing at an intermediary node of the (ML) model based on the output of the intermediary node. Based on the output of the intermediary node satisfying a condition, the device may use the output of the intermediary node to generate an output responsive to the input. Conversely, if the output of the intermediary node does not satisfy a condition, the device may send the output of the intermediary node to the remote system, so the remote system can use another machine learning model to complete processing with respect to the input.

Multi-device speech processing

Growing community of inventors

Waltham, MA, United States of America

Rahul Gupta