Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for index-side synonym expansion. One method includes obtaining a token sequence for a resource and indexing a token in the token sequence. The indexing includes applying one or more stemming rules to the particular token to generate a stemmed form of the token, obtaining a variant of the stemmed form of the token, and storing data associating the resource with both the token and the variant as index terms for the resource in a search engine index.

Abstract: Personalized network searching, in which a search query is received from a user, and a request is received to personalize a search result. Responsive to the search query and the request to personalize the search result, a personalized search result is generated by searching a personalized search object. Responsive to the search query, a general search result is generated by searching the general search object. The personalized search result and the general search result are provided to a client device, an advertisement is selected based at least in part upon the personalized search object, and the advertisement, the personalized search result, and the general search result are displayed.

Abstract: The present disclosure provides projection neural networks and example applications thereof. In particular, the present disclosure provides a number of different architectures for projection neural networks, including two example architectures which can be referred to as: Self-Governing Neural Networks (SGNNs) and Projection Sequence Networks (ProSeqoNets). Each projection neural network can include one or more projection layers that project an input into a different space. For example, each projection layer can use a set of projection functions to project the input into a bit-space, thereby greatly reducing the dimensionality of the input and enabling computation with lower resource usage. As such, the projection neural networks provided herein are highly useful for on-device inference in resource-constrained devices. For example, the provided SGNN and ProSeqoNet architectures are particularly beneficial for on-device inference such as, for example, solving natural language understanding tasks on-device.

Abstract: Systems and methods for providing a personalized visual display of multiple products are provided. A described method includes receiving product information for a set of multiple products and user information for a particular user or user device and selecting a plurality of the multiple products estimated to be most relevant to the particular user or user device based on the product information for the set of multiple products and the user information for the particular user or user device. The method further includes generating a personalized visual display of the selected products including product images for multiple of the selected products. All of the products in the personalized visual display may be associated with the same content provider. The method further includes causing the personalized visual display to be presented via the user device.

Abstract: Techniques are described herein for training and/or utilizing a query canonicalization system. In various implementations, a query canonicalization system can include a classification model and a canonicalization model. A classification model can be used to determine if a search query is well-formed. Additionally or alternatively, a canonicalization model can be used to determine a well-formed variant of a search query in response to determining a search query is not well-formed. In various implementations, a canonicalization model portion of a query canonicalization system can be a sequence to sequence model.

Abstract: Secure communication in mobile digital pages is provided. The system receives an electronic document and validates the electronic document for storage in a cache server. The system receives a request for the electronic document and provides it to a viewer component on a client computing device. The viewer component loads the electronic document in an iframe. The viewer component executes a runtime component to receive, via a secure communication channel, a tag from the electronic document. The system receives the tag and selects a data value for transmission to the viewer component. The viewer components provides the data value to cause the runtime component to execute an action with the data value.

Abstract: Methods, apparatus, systems, and computer-readable media are provided for allowing an automated assistant routine to be interrupted during performance of the routine. A routine can correspond to a set of actions to be performed at the direction of the automated assistant. When the routine is initialized and a user subsequently issues a command to interrupt the routine, the automated assistant can modify a status identifier for the routine. That status identifier can be stored at a database and allow other applications and/or devices that are operating to complete the routine to be put on notice that the user has requested the routine be interrupted. The database can be accessible to one or more devices and/or applications, such as third party applications, in order to provide a medium through which the devices and/or applications can check the statuses of various automated assistant routines.

Abstract: Dynamically controlling output from a device, such as an automated assistant device. Control of the output can be based on, for example, a condition and/or physiological attribute(s) of a user of the device. Various implementations dynamically control the output to improve sleep quality for the user and/or mitigate waste of computational and/or network resources.

Abstract: Techniques are described herein for enabling the use of “dynamic” or “context-specific” hot words for an automated assistant. In various implementations, an automated assistant may be operated at least in part on a computing device. Audio data captured by a microphone may be monitored for default hot word(s). Detection of one or more of the default hot words may trigger transition of the automated assistant from a limited hot word listening state into a speech recognition state. Transition of the computing device into a given state may be detected, and in response, the audio data captured by the microphone may be monitored for context-specific hot word(s), in addition to or instead of the default hot word(s). Detection of the context-specific hot word(s) may trigger the automated assistant to perform a responsive action associated with the given state, without requiring detection of default hot word(s).

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating, in response to a single input operating system command that is invoked independent of a native application, a message that includes an image of a particular environment instance of the native application that was displayed when the single input operating system command and a uniform resource identifier of the particular environment instance of the native application.

Abstract: According to an aspect, a method for multi-channel echo cancellation includes receiving a microphone signal and a multi-channel loudspeaker driving signal. The multi-channel loudspeaker driving signal includes a first driving signal that drives a first loudspeaker, and a second driving signal that drives a second loudspeaker. The first driving signal is substantially the same as second driving signal. The microphone signal includes a near-end signal with echo. The method includes determining a unique solution for acoustic transfer functions for a present acoustic scenario based on the microphone signal and the multi-channel loudspeaker driving signal. The acoustic transfer functions include first and second acoustic transfer function. The unique solution is determined based on time-frequency transforms of observations from the present acoustic scenario and at least one previous acoustic scenario.

Abstract: Techniques are described for selectively adapting and/or selectively utilizing a noise reduction technique in detection of one or more features of a stream of audio data frames. For example, various techniques are directed to selectively adapting and/or utilizing a noise reduction technique in detection of an invocation phrase in a stream of audio data frames, detection of voice characteristics in a stream of audio data frames (e.g., for speaker identification), etc. Utilization of described techniques can result in more robust and/or more accurate detections of features of a stream of audio data frames in various situations, such as in environments with strong background noise. In various implementations, described techniques are implemented in combination with an automated assistant, and feature(s) detected utilizing techniques described herein are utilized to adapt the functionality of the automated assistant.

Abstract: Analyzing or debugging applications is provided. The system identifies an action for an application provided by a developer. The system determines a first classification score based on historical execution of the action. The system generates a machine generated action for the application based on metadata associated with the application. The system determines a second classification score based on a comparison of the action with the machine generated action. The system selects, via a matching program, a second application that matches the application. The system determines a third classification score based on a comparison of an action approved for the second application with the action provided by the application developer. The system updates a delivery control parameter based on the first classification score, the second classification score and the third classification score. The system controls delivery of the application based on the delivery control parameter.

Abstract: Systems and techniques are provided for anomalous path detection within cameras' fields of view. Video of the field of view of a camera in an environment may be received from the camera. A person may be detected in the video. Motion of the person in the video may be tracked to generate a motion path. Contextual data for the motion path may be received. The motion path and contextual data may be stored in a camera training data set. A camera model for the camera and the field of view may be generated by inputting the camera training data set to a machine learning system. The camera model for the camera and the field of view may be stored.

Abstract: A response to a first query related to a geographic area, and search results responsive to a second query related to the geographic area, are received from a server system. A digital map of the geographic is provided via a user interface. A focus is provided to the first query to illustrate the response on the digital map, where the response defines a geographic context. A subset of the search results is identified based on a relationship to the geographic context. The search results in the identified subset are visually emphasized on the digital map relative to the rest of the search results.

Abstract: Adaptive navigation techniques are disclosed that allow navigation systems to learn from a user's personal driving history. As a user drives, models are developed and maintained to learn or otherwise capture the driver's personal driving habits and preferences. Example models include road speed, hazard, favored route, and disfavored route models. Other attributes can be used as well, whether based on the user's personal driving data or driving data aggregated from a number of users. The models can be learned under explicit conditions (e.g., time of day/week, driver ID) and/or under implicit conditions (e.g., weather, drivers urgency, as inferred from sensor data). Thus, models for a plurality of attributes can be learned, as well as one or more models for each attribute under a plurality of conditions. Attributes can be weighted according to user preference. The attribute weights and/or models can be used in selecting a best route for user.

Abstract: Techniques described herein enable a user to interact with an automated assistant and obtain relevant output from the automated assistant without requiring arduous typed input to be provided by the user and/or without requiring the user to provide spoken input that could cause privacy concerns (e.g., if other individuals are nearby). The assistant application can operate in multiple different image conversation modes in which the assistant application is responsive to various objects in a field of view of the camera. The image conversation modes can be suggested to the user when a particular object is detected in the field of view of the camera. When the user selects an image conversation mode, the assistant application can thereafter provide output, for presentation, that is based on the selected image conversation mode and that is based on object(s) captured by image(s) of the camera.

Abstract: Automatically generating and/or automatically transmitting a status of a user. The status is transmitted for presentation to one or more additional users via corresponding computing device(s) of the additional user(s). Some implementations are directed to determining both: a status of a user, and a predicted duration of that status; and generating a status notification that includes the status and that indicates the predicted duration. Some implementations are additionally or alternatively directed to utilizing at least one trust criterion in determining whether to provide a status notification of a user to an additional user and/or in determining what status notification to provide to the additional user. Some implementations are additionally or alternatively directed to training and/or use of machine learning model(s) in determining a status of a user and/or a predicted duration of that status.

Abstract: A mobile image capture device is provided that can include an image capture system operable to capture image frames and be configured to: provide a live video stream for display in a viewfinder portion of a user interface that depicts at least a portion of a current field of view of the image capture system; store a video segment from the live video stream in a temporary image buffer that includes a plurality of image frames captured by the image capture system; receive a user input that is directed to the viewfinder portion of the user interface that requests a rewind operation; and, in response to such user input, perform the rewind operation in the viewfinder portion of the user interface with respect to the video segment in which at least two of the image frames of the video segment can be provided for display in the user interface in a reverse chronological order.