Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: This document relates to automated analysis of images. One example method involves obtaining an image and text associated with the image, detecting two or more objects in the image, and determining respective locations of the two or more detected objects in the image. The example method also involves determining whether a spatial relationship between the two or more detected objects matches a corresponding spatial relationship expressed by the text based at least on the respective locations of the two or more detected objects. The example method also involves outputting a value reflecting whether the spatial relationship between the two or more detected objects matches the corresponding spatial relationship expressed by the text.

Abstract: This document relates to automated analysis of images. One example method involves obtaining an image and text associated with the image, detecting two or more objects in the image, and determining respective locations of the two or more detected objects in the image. The example method also involves determining whether a spatial relationship between the two or more detected objects matches a corresponding spatial relationship expressed by the text based at least on the respective locations of the two or more detected objects. The example method also involves outputting a value reflecting whether the spatial relationship between the two or more detected objects matches the corresponding spatial relationship expressed by the text.

Abstract: A computing system is provided that includes processing circuitry and associated memory. The processing circuitry is configured to implement a program using portions of the associated memory, to receive, via an edit operation, digital content and provenance metadata associated with the digital content. The processing circuitry implementing the program is further configured to determine, via a provenance determination module, that a textual portion of the digital content is model-generated and originated from a generative model, based on the provenance metadata, and output the digital content to a graphical user interface with a visual indication that the textual portion of the digital content is model-generated.

Abstract: A computing system for conditional generation of protein sequences includes processing circuitry that implements a denoising diffusion probabilistic model. In an inference phase, the processing circuitry receives an instruction to generate a predicted protein sequence having a target functionality, the instruction including first conditional information and second conditional information. The processing circuitry concatenates a first conditional information embedding generated by a first encoder and a second conditional information embedding generated by a second encoder to produce a concatenated conditional information embedding. The processing circuitry samples noise from a distribution function and combines the concatenated conditional information embedding with the sampled noise to produce a noisy concatenated input.

Abstract: The technology described herein builds an optimal refresh schedule by minimizing a cost function constrained by an available refresh bandwidth. The cost function receives an importance score for a content item and a change rate for the content item as input in order to optimize the schedule. The cost function is considered optimized when a refresh schedule is found that minimizes the cost while using the available bandwidth and no more. The technology can build an optimized schedule to refresh content with incomplete change data, content with complete change data, or a mixture of content with and without complete change data. It can also re-learn content item change rates from its own schedule execution history and re-compute the refresh schedule, ensuring that this schedule takes into account the latest trends in content item updates.

Abstract: Systems and methods to determine when a media is a high-fidelity reproduction of an original media from a trusted entity are disclosed. In certain aspects, systems and method for generating a fragile watermark are disclosed. The fragile watermark may be inserted into digital media in a manner such that the watermark cannot be identified if the media content is significantly altered. Media content may be subsequently analyzed to determine the presence of a fragile watermark. When the fragile watermark is present, provenance of the media content can be verified and an indication of provenance is provided to the user.

Abstract: Systems and methods to determine when a media is a high-fidelity reproduction of an original media from a trusted entity are disclosed. In certain aspects, systems and method for generating a fragile watermark are disclosed. The fragile watermark may be inserted into digital media in a manner such that the watermark cannot be identified if the media content is significantly altered. Media content may be subsequently analyzed to determine the presence of a fragile watermark. When the fragile watermark is present, provenance of the media content can be verified and an indication of provenance is provided to the user.

Abstract: This document relates to automated analysis of images. One example method involves obtaining an image and text associated with the image, detecting two or more objects in the image, and determining respective locations of the two or more detected objects in the image. The example method also involves determining whether a spatial relationship between the two or more detected objects matches a corresponding spatial relationship expressed by the text based at least on the respective locations of the two or more detected objects. The example method also involves outputting a value reflecting whether the spatial relationship between the two or more detected objects matches the corresponding spatial relationship expressed by the text.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.

Abstract: The technology described herein builds an optimal refresh schedule by minimizing a cost function constrained by an available refresh bandwidth. The cost function receives an importance score for a content item and a change rate for the content item as input in order to optimize the schedule. The cost function is considered optimized when a refresh schedule is found that minimizes the cost while using the available bandwidth and no more. The technology can build an optimized schedule to refresh content with incomplete change data, content with complete change data, or a mixture of content with and without complete change data. It can also re-learn content item change rates from its own schedule execution history and re-compute the refresh schedule, ensuring that this schedule takes into account the latest trends in content item updates.

Abstract: The technology described herein builds an optimal refresh schedule by minimizing a cost function constrained by an available refresh bandwidth. The cost function receives an importance score for a content item and a change rate for the content item as input in order to optimize the schedule. The cost function is considered optimized when a refresh schedule is found that minimizes the cost while using the available bandwidth and no more. The technology can build an optimized schedule to refresh content with incomplete change data, content with complete change data, or a mixture of content with and without complete change data. It can also re-learn content item change rates from its own schedule execution history and re-compute the refresh schedule, ensuring that this schedule takes into account the latest trends in content item updates.

Abstract: The subject disclosure is directed towards processing evidence, which may include high-dimensional streaming evidence, into a future belief state. The existing evidence is used to project a belief about a future state. The future belief state may be used to determine whether to wait for additional evidence, or to act now without waiting for additional evidence, e.g., based on a cost of the delay. For example, an autonomous assistant may decide based upon the belief whether to engage a person or not, or to wait for more information before the engagement decision is made.

Abstract: A system is described for integrating a search engine and one or more social network resources. The system operates by determining whether a search operation being conducted by a user warrants interaction with a social network resource. If so, the system may provide an invitation to the user to forward a query-related message to the social network resource. The system then sends the message to a group of contacts via the social network resource, where the group of contacts can be defined in various ways. The system receives a response from at least one contact in the group of contacts and presents that response to the user using various delivery mechanisms, as governed by various delivery timings.

Abstract: Systems and methods to determine when a media is a high-fidelity reproduction of an original media from a trusted entity are disclosed. In certain aspects, systems and method for generating a fragile watermark are disclosed. The fragile watermark may be inserted into digital media in a manner such that the watermark cannot be identified if the media content is significantly altered. Media content may be subsequently analyzed to determine the presence of a fragile watermark. When the fragile watermark is present, provenance of the media content can be verified and an indication of provenance is provided to the user.

Abstract: The technology described herein builds an optimal refresh schedule by minimizing a cost function constrained by an available refresh bandwidth. The cost function receives an importance score for a content item and a change rate for the content item as input in order to optimize the schedule. The cost function is considered optimized when a refresh schedule is found that minimizes the cost while using the available bandwidth and no more. The technology can build an optimized schedule to refresh content with incomplete change data, content with complete change data, or a mixture of content with and without complete change data. It can also re-learn content item change rates from its own schedule execution history and re-compute the refresh schedule, ensuring that this schedule takes into account the latest trends in content item updates.

Abstract: The claimed subject matter provides systems and/or methods that facilitate inferring probability distributions over the destinations and/or routes of a user, from observations about context and partial trajectories of a trip. Destinations of a trip are based on at least one of a prior and a likelihood based at least in part on the received input data. The destination estimator component can use one or more of a personal destinations prior, time of day and day of week, a ground cover prior, driving efficiency associated with candidate locations, and a trip time likelihood to probabilistically predict the destination. In addition, data gathered from a population about the likelihood of visiting previously unvisited locations and the spatial configuration of such locations may be used to enhance the predictions of destinations and routes.

Abstract: This document relates to automated generation of machine learning models, such as neural networks. One example system includes a hardware processing unit and a storage resource. The storage resource can store computer-readable instructions cause the hardware processing unit to perform an iterative model-growing process that involves modifying parent models to obtain child models. The iterative model-growing process can also include selecting candidate layers to include in the child models based at least on weights learned in an initialization process of the candidate layers. The system can also output a final model selected from the child models.