Abstract: Implementations relate to determining a language for speech recognition of a spoken utterance, received via an automated assistant interface, for interacting with an automated assistant. Implementations can enable multilingual interaction with the automated assistant, without necessitating a user explicitly designate a language to be utilized for each interaction. Selection of a speech recognition model for a particular language can based on one or more interaction characteristics exhibited during a dialog session between a user and an automated assistant. Such interaction characteristics can include anticipated user input types, anticipated user input durations, a duration for monitoring for a user response, and/or an actual duration of a provided user response.

Abstract: Video streams uploaded to a video hosting platform are transcoded using quality-normalized transcoding parameters dynamically selected using a learning model. Video frames of a video stream are processed using the learning model to determine bitrate and quality score pairs for some or all possible transcoding resolutions. The listing of bitrate and quality score pairs determined for each resolution is processed to determine a set of transcoding parameters for transcoding the video stream into each resolution. The bitrate and quality score pairs of a given listing may be processed using one or more predefined thresholds, which may, in some cases, refer to a weighted distribution of resolutions according to watch times of videos of the video hosting platform. The video stream is then transcoded into the various resolutions using the set of transcoding parameters selected for each resolution.

Abstract: Implementations disclosed herein are directed to utilizing ephemeral learning techniques and/or federated learning techniques to update audio-based machine learning (ML) model(s) based on processing streams of audio data generated via radio station(s) across the world. This enables the audio-based ML model(s) to learn representations and/or understand languages across the world, including tail languages for which there is no/minimal audio data. In various implementations, one or more deduping techniques may be utilized to ensure the same stream of audio data is not overutilized in updating the audio-based ML model(s). In various implementations, a given client device may determine whether to employ an ephemeral learning technique or a federated learning technique based on, for instance, a connection status with a remote system. Generally, the streams of audio data are received at client devices, but the ephemeral learning techniques may be implemented at the client device and/or at the remote system.

Abstract: Techniques are described herein for selectively processing a user's utterances captured prior to and after an event that invokes an automated assistant to determine the user's intent and/or any parameters required for resolving the user's intent. In various implementations, respective measures of fitness for triggering responsive action by the automated assistant may be determined for pre-event and a post-event input streams. Based on the respective measures of fitness, one or both of the pre-event input stream or post-event input stream may be selected and used to cause the automated assistant to perform one or more responsive actions.

Abstract: The present disclosure provides systems and methods that perform face reconstruction based on an image of a face. In particular, one example system of the present disclosure combines a machine-learned image recognition model with a face modeler that uses a morphable model of a human's facial appearance. The image recognition model can be a deep learning model that generates an embedding in response to receipt of an image (e.g., an uncontrolled image of a face). The example system can further include a small, lightweight, translation model structurally positioned between the image recognition model and the face modeler. The translation model can be a machine-learned model that is trained to receive the embedding generated by the image recognition model and, in response, output a plurality of facial modeling parameter values usable by the face modeler to generate a model of the face.

Abstract: A mapping system allows a user to interact with any location on a digital map and present the user with location related information associated with the selected location. The location related information may be in the form or a card, pop-up, image, or other graphic and may be displayed on the map at or near the selected location, around the map, etc. The displayed location related information may include predetermined or pre-stored data about the location or may include location related information collected and generated on the fly in response to the user interaction with the digital map. The displayed location related information may be displayed in the same graphical format on the digital map regardless of whether location related information is predetermined information about the location that already exists or information that is collected and generated on the fly in response to the user interaction.

Abstract: Smart glasses including a first audio device, a second audio device, a frame including a first portion, a second portion, and a third portion, the second portion and the third portion are moveable in relation to the first portion, the second portion including the first audio device and the third portion including the second audio device, and a processor configured to cause the first audio device to generate a signal, receive the signal via the second audio device, estimate a distance based on the received signal, and determine a configuration of the frame.

Abstract: A sensor system includes a touch sensor having a plurality of conductive sensing elements integrated with a flexible substrate. A first subset of sensing elements is coupled to a first side of the flexible substrate and a second subset of sensing elements is coupled to a second side of the flexible substrate. At least one sensing element of the first subset overlies a second sensing element and a third sensing element of the second subset. The sensor system is configured to obtain touch data based at least in part on a respective response to the touch input by the plurality of conductive sensing elements. The sensor system is configured to determine whether the touch input is associated with the first subset of conductive sensing elements or the second subset of conductive sensing elements based at least in part on the respective response to the touch input by the sensing elements.

Abstract: To determine whether a user is authorized to make a particular audio request during navigation, a client device receives a request for navigation directions from a starting location to a destination location. The client device provides a set of navigation directions for traversing to the destination location along a route. During a navigation session, an audio request related to the route is received from a user. The client device determines an authorization level of the user based on the audio request, and provides a response to the request based on the authorization level of the user.

Abstract: A computing device may implement a method for creating and providing reminders during navigation sessions. The method may include initiating a navigation session for providing a set of navigation instructions to a user from a starting location to a destination location along a route, and receiving, from a user during the navigation session, an input including a request to create a reminder. The method may further include analyzing the input to determine: (i) a content of the reminder, and (ii) a trigger condition for providing the reminder, the trigger condition including a trigger location on the route, and storing the reminder, including the content and the trigger condition. The method may also include, detecting, during a subsequent navigation session, the trigger condition, and, in response to detecting the trigger condition, providing the reminder.

Abstract: A computer-implemented method for providing dynamically scrolling content within a vertically scrolling online publication is implemented by a content serving computing device in communication with a memory. The method includes retrieving a dynamically scrolling content item, receiving a set of display information related to a user computing device, and serving the dynamically scrolling content item within a slot of a vertically scrolling online publication presented at the user computing device, wherein the dynamically scrolling content item is served based partially on the set of display information, wherein the dynamically scrolling content item is configured to scroll horizontally in response to a first haptic input received by a user of the user computing device.

Abstract: This disclosure is directed to a quantum computing system (QCS) that includes a cryogenic sub-system, a signal-generating element, a first signal-splitting element, a first transmission path, a second transmission path, a third transmission path, and a quantum device. A first environment is located outside the cryogenic sub-system and a second environment is associated with the cryogenic sub-system. The signal-generating element generates a first signal. The first signal-splitting element is positioned within the second environment. The quantum device is positioned within the cryogenic sub-system. The first transmission path transmits the first signal from the signal-generating element to the first signal-splitting element. The first signal-splitting element subdivides the first signal into a second signal and a third signal. The third transmission transmits the third signal from the first signal-splitting element to the first environment.

Abstract: Systems and techniques include using a sensor of a computing device to detect the presence of a first portion of a code, the code including at least the first portion and a second portion, where the first portion of the code is decodable and includes an identifier and the second portion of the code is non-decodable. The computing device recognizes the identifier in the first portion of the code and obtains instructions for decoding the second portion of the code using the identifier and/or data associated with the identifier. The instructions to decode the second portion of the code are processed to generate a decoded second portion of the code. The computing device performs an action defined in the decoded second portion of the code.

Abstract: A computer-implemented method that provides watermark-based image reconstruction to compensate for lossy encoding schemes. The method can generate a difference image describing the data loss associated with encoding an image using a lossy encoding scheme. The difference image can be encoded as a message and embedded in the encoded image using a watermark and later extracted from the encoded image. The difference image can be added to the encoded image to reconstruct the original image. As an example, an input image encoded using a lossy JPEG compression scheme can be embedded with the lost data and later reconstructed, using the embedded data, to a fidelity level that is identical or substantially similar to the original.

Abstract: Proactive caching, at a client device (e.g., a feature phone), of transient assistant action suggestions for selective rendering by an assistant client application of the client device. A transient assistant action suggestion, when rendered via an assistant client application and selected, causes the assistant client application to initiate performance of a corresponding assistant action. In various implementations, a prefetched transient action suggestion can be a time-constrained suggestion that includes at least associated rendering restriction metadata that defines one or more temporal windows to which rendering of the time-constrained suggestion is restricted. Proactive cache refresh rate metadata can also be associated with transient action suggestion(s) and defines a duration during which the assistant client application is to refrain from interfacing with a remote system to prefetch updated transient assistant action suggestions.

Abstract: Techniques are described herein for using a smart device such as a standalone assistant-centric interactive speaker and/or a standalone assistant-centric interactive display with speaker(s) to send a message using a messaging application on a client device such as a smartphone. A method includes: receiving, by a first device, a request from a first user to send a message to a second user; determining that a messaging application corresponding to the request is unavailable on the first device; and in response to determining that the messaging application corresponding to the request is unavailable on the first device: selecting a second device on which the messaging application corresponding to the request is available; and sending, to the second device, a command that causes the second device to send the message from the first user to the second user using the messaging application on the second device.

Abstract: Techniques for preparing datasets for geo experiments and improving accuracy of geo experiments are presented herein. The system can access a dataset of a plurality of geographic pairs. Additionally, the system can calculate a first outcome estimate based on a difference in response data and a difference in input data for a first geographic pair. Moreover, the system can calculate a plurality of experimental uncertainty estimates associated with the plurality of geographic pairs during an experimental time interval. The system can access historical data associated with the plurality of geographic pairs. Furthermore, the system can determine a beta value and a trim rate that reduces a sum of the plurality estimates. Subsequently, the system can remove, based on the first outcome estimate and the beta value, the first geographic pair from the plurality of geographic pairs to generate the first subset of geographic pairs.

Abstract: The present disclosure provides a statistical, articulated 3D human shape modeling pipeline within a fully trainable, modular, deep learning framework. In particular, aspects of the present disclosure are directed to a machine-learned 3D human shape model with at least facial and body shape components that are jointly trained end-to-end on a set of training data. Joint training of the model components (e.g., including both facial, hands, and rest of body components) enables improved consistency of synthesis between the generated face and body shapes.

Abstract: An encoding process performed by a computing device (e.g., a user's private device) can include obtaining private data that includes a private value. According to an aspect of the present disclosure, the computing device can produce a plurality of messages that respectively comprise a plurality of message values, where a total sum of the plurality of message values approximates the private value, and where at least one of the plurality of message values is randomly selected. The device can provide the plurality of messages for aggregation with a plurality of additional messages respectively generated for a plurality of additional private values. For example, the messages can be transmitted to a shuffler model configured to shuffle the plurality of messages with the plurality of additional messages.

Abstract: A computer-implemented method for generating geocoded user information is disclosed. The method comprises searching user data across multiple different data corpuses for entries having location-related information and determining locations for the location-related information. The method further comprises generating a map showing a current location of a mobile device along with representations of the entries having location-related information, at the determined locations, for entries from the multiple different data corpuses.