Abstract: In an embodiment, a computing system can include one or more processors and one or more non-transitory computer-readable media that store instructions that, when executed by the one or more processors, cause the computing system to perform operations. The operations can include: receiving an internal encoder state of an encoder running on a first computing device being used to participate in a video conference currently in progress; receiving data indicative of a second computing device being used to join the video conference; compressing, based at least in part on receipt of the data, the internal encoder state to generate a compressed internal encoder state of the encoder; and/or transmitting the compressed internal encoder state to the second computing device to synchronize the internal encoder state of the encoder running on the first computing device with an internal decoder state of a decoder running on the second computing device.

Abstract: Systems and methods for searching using machine-learned model-generated outputs can provide a user with a medium for generating a theoretical dataset that can then be matched to a real world example. The systems and methods can include selecting a plurality of terms, which can be utilized to generate a prompt input that can be processed by a dataset generation model to generate a plurality of model-generated datasets. A selection can then be received that selects a particular model-generated database to utilize to query a database.

Abstract: A method of sharing and synchronizing a plurality of electronically stored resources between a cloud file system and a client file system is provided. The method includes receiving one or more change events wherein each change event indicates an independent change to one or more electronically stored files in a first client file system, holding the received change events for a period of time, and determining whether any of the held change events are combinable with other held change events indicating one or more independent changes to one or more electronically stored files in one or more client file systems comprising the first client file system.

Abstract: Techniques and apparatuses are described for enabling a base station to enable peer-to-peer communication among multiple user equipment (UE) over a mmWave link. The techniques described herein overcome challenges that the multiple UEs might otherwise face in trying to establish peer-to-peer links on their own. By relying on the base station to grant air interface resources for the UE to perform peer-to-peer communications with the UE, the UE can communicate directly with the other UE, independent of links that the UE or the other UE maintains with the base station. Furthermore, reliance on the base station may help the UE and the other UE mitigate interference from other nearby mmWave links that are separate from the peer-to-peer wave link. In addition, by relying on the base station to specify the beam sweeping pattern, beam acquisition by the UE and the other UE may be improved.

Abstract: Generating granular clusters for real-time processing is provided. The systems can identify tokens based on aggregating input from computing devices over a time interval. The systems can identify, based on metrics, a subset of tokens for cluster generation. The systems can generate, via a clustering technique, token clusters from the subset of the tokens, each of the token clusters comprising two or more tokens from the subset of the tokens. The systems can apply a de-duplication technique to each of the token clusters. The systems can apply a filtering technique to the token clusters to remove tokens erroneously grouped in a token cluster. The systems can assign, based on a selection process, a label for each of the token clusters. The systems can activate, based on a number of remaining tokens in each of the token clusters, a subset of the token clusters for real-time content selection.

Abstract: This document describes a security camera with an angled cable attachment for an increased downward viewing angle. The security camera is battery-powered and can be magnetically coupled to a mounting device and electrically connected to another device via a cable. The cable has a cable attachment that, when coupled to the security camera, is angled toward a front of the security camera. This cable angle enables an increased downward tilt angle of the security camera by reducing interference of the cable attachment with the mounting device when the security camera is tilted downward. The security camera also has exposed contacts on a printed circuit board that connect with pins on the cable attachment.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing sequence modeling tasks using insertions. One of the methods includes receiving a system input that includes one or more source elements from a source sequence and zero or more target elements from a target sequence, wherein each source element is selected from a vocabulary of source elements and wherein each target element is selected from a vocabulary of target elements; generating a partial concatenated sequence that includes the one or more source elements from the source sequence and the zero or more target elements from the target sequence, wherein the source and target elements arranged in the partial concatenated sequence according to a combined order; and generating a final concatenated sequence that includes a finalized source sequence and a finalized target sequence, wherein the finalized target sequence includes one or more target elements.

Abstract: An apparatus includes a first group of superconducting cavities and a second group of superconducting cavities, each of which is configured to receive multiple photons. The apparatus includes couplers, where each coupler couples one superconducting cavity from the first group with one cavity from the second group such that the photons in the coupled superconducting cavities interact. A first superconducting cavity of the first group is connected to a second superconducting cavity of the second group, such that photons of the first and second superconducting cavities are shared by each of the first and second superconducting cavities. The first superconducting cavity is coupled to at least one other superconducting cavity of the first group to which the second superconducting cavities are coupled, and the second superconducting cavity is coupled to at least one other superconducting cavity of the second group to which the first superconducting cavities are coupled.

Abstract: Apparatus and methods for performing qubit readout. In one aspect, an apparatus includes a qubit that operates at a qubit frequency; a frequency controller that is configured to control the qubit frequency and that during a qubit measurement operation is configured to: determine a compensation pulse that when applied to the qubit, counteracts qubit frequency changes during the qubit measurement operation; and apply the determined compensation pulse to the qubit during the qubit measurement operation to maintain the qubit frequency.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for receiving graph data representing an input graph comprising a plurality of vertices connected by edges; generating, from the graph data, vertex input data representing characteristics of each vertex in the input graph and pair input data representing characteristics of pairs of vertices in the input graph; and generating order-invariant features of the input graph using a neural network, wherein the neural network comprises: a first subnetwork configured to generate a first alternative representation of the vertex input data and a first alternative representation of the pair input data from the vertex input data and the pair input data; and a combining layer configured to receive an input alternative representation and to process the input alternative representation to generate the order-invariant features.

Abstract: A hardware accelerator can store, in multiple memory storage areas in one or more memories on the accelerator, input data for each processing time step of multiple processing time steps for processing sequential inputs to a machine learning model. For each processing time step, the following is performed. The accelerator can access a current value of a counter stored in a register within the accelerator to identify the processing time step. The accelerator can determine, based on the current value of the counter, one or more memory storage areas that store the input data for the processing time step. The accelerator can facilitate access of the input data for the processing time step from the one or more memory storage areas to at least one processor coupled to the one or more memory storage areas. The accelerator can increment the current value of the counter stored in the register.

Abstract: The present invention pertains to enhancement or refinement of estimated locations based upon user-specific information. Upon user authorization, geographical information is extracted from a number of user-related sources, including the web browser history, search history, maps history, address book, e-mail archives and calendar entries. Such information is used to build a spatial index of specific physical locations for a geocoded result set. From this, heat maps identifying particular locations from the user-related sources are created for different periods of time. The heat maps may be used to refine an initial location estimate of the user. This may be done by determining whether one or more positions in a given heat map provide a more accurate position of the user than the initial estimate. If so, a best position is selected. This can be used to provide enhanced driving directions to the user.

Abstract: Described techniques enable thwart detection for warning lights associated with recording devices, such as cameras. A warning light may be used to alert bystanders that a recording is in progress, but an operator of a recording device may attempt to circumvent the warning light by obstructing the warning light. A lightpipe or other light guiding element may be used to guide a redirection of the warning light to a light sensor. Then, when the redirection exceeds a threshold, which may be determined based in part on a baseline level of ambient light, recording operations of the recording device may be automatically deactivated.

Abstract: A method includes receiving one or more past trajectories navigated by a robotic device in an environment, wherein the one or more past trajectories are associated with initial environmental sensor data and one or more obstacle detection heuristics. The method also includes determining, based at least on subsequent environmental sensor data, one or more updated obstacle detection heuristics. The method further includes determining, based on the one or more updated obstacle detection heuristics and the initial environmental sensor data, one or more predicted drivable areas in the environment. The method additionally includes, based on the one or more predicted drivable areas including the one or more past trajectories, using the one or more updated obstacle detection heuristics to determine future navigation of the robotic device.

Abstract: An apparatus includes a substrate, a classical computing processor formed on the substrate, a quantum computing processor formed on the substrate, and one or more coupling components between the classical computing processor and the quantum computing processor, the one or more coupling components being formed on the substrate and being configured to allow data exchange between the classical computing processor and the quantum computing processor.

Abstract: A game streaming system encodes a raw frame to generate an encoded frame for a game stream including a plurality of frames for display. The system identifies a first encoding artifact based on a first difference between the encoded frame and the raw frame, and applies an overlay to the encoded frame. The overlay includes a first visual indicator indicating the first encoding artifact.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for evaluating candidate output sequences using language model neural networks. In particular, an auto-regressive language model neural network is used to generate a candidate output sequence. The same auto-regressive language model neural network is used to evaluate the candidate output sequence to determine rating scores for each of one or more criteria. The rating score(s) are then used to determine whether to provide the candidate output sequence.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for dialogue systems. A transcription of a user utterance is obtained. The transcription of the utterance is tokenized to identify multiple tokens for the utterance. Token-level utterance encodings corresponding to different tokens of the transcription are generated. A system action encoding from data indicating system actions previously performed by the dialogue system are generated. A dialogue context vector based on the utterance encoding and the system action encoding are generated. The token-level utterance encodings, the system action encoding, and the dialogue context vector are processed using a slot tagger to produce token-level output vectors. A limited set of candidate token classifications for the tokens of the user utterance are determined based on the token-level utterance encodings. A response for output is provided in response to the user utterance.

Abstract: A method includes, for each training sample of a plurality of training samples, processing, using an RNN-T model, a corresponding sequence of acoustic frames to obtain an n-best list of speech recognition hypotheses, and, for each speech recognition hypothesis of the n-best list, determining a corresponding number of word errors relative to a corresponding ground-truth transcription. For a top-ranked hypothesis from the n-best list, the method includes determining a first loss based on the corresponding ground-truth transcription. The method includes identifying, as an oracle hypothesis, the speech recognition hypothesis from the n-best list having the smallest corresponding number of word errors relative to the corresponding ground-truth transcription, and determining a second loss for the oracle hypothesis based on the corresponding ground-truth transcription.

Abstract: A method includes receiving a sequence of acoustic frames characterizing an utterance. During a first pass, the method includes generating first-pass audio encodings based on the sequence of acoustic frames using a stack of mask-conformer blocks of an acoustic encoder, generating a first-pass transcription of the utterance based on the first-pass audio encodings using a speech recognition decoder, and generating a first-pass masked output sequence using a mask-predict decoder of the acoustic encoder. During a second pass, the method includes generating second-pass audio encodings by performing cross-attention on the sequence of acoustic frames and the masked first-pass transcription using the stack of mask-conformer blocks of the acoustic encoder and generating a second-pass transcription of the utterance based on the second-pass audio encodings using the speech recognition decoder.