Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for in-process adjustment to crushing systems.

Abstract: Implementations are described herein for automatically generating multimodal geospatial workflows for accomplishing geospatial tasks. In various implementations, a natural language request may be processed based on generative model(s) such as LLM(s) to generate workflow output tokens that identify high-level actions for completing a geospatial task conveyed in the natural language request. First data indicative of the high-level actions may be processed using one or more of the generative models to generate dataset output tokens that identify responsive dataset(s) that likely contain data responsive to the geospatial task. Second data indicative of both the high-level actions and the responsive dataset(s) may be processed based on one or more of the generative models to generate data manipulation output tokens that identify data manipulation instructions for assembling data from the responsive dataset(s) into a response that fulfills the geospatial task.

Abstract: This disclosure describes a system, method, and non-transitory computer-readable medium for image search-based object detection of utility assets in image databases. The method includes receiving an input image of a utility asset and a query bounding box representing an image-based object query. Bounding boxes of objects represented in the input image are generated based on the input image and the query bounding box, in which anchoring boxes corresponding to object classifications are identified from the bounding boxes. A textual label is determined for a selected subset of anchoring boxes. An image embedding representing the region is encoded, and image tokens are generated based on the encoded image embedding. Output images of other utility assets relevant to the image-based object query are identified from images in an image database, based on at least one of (i) the encoded image embedding, (ii) the image tokens, or (iii) the textual label. The output images are provided for output.

Abstract: Methods, systems, and apparatus for providing a ML model for inference, the ML model having been trained using a first set of training data to provide predictions associated with an adverse event, after training of the ML model, receiving data from one or more data sources, the data representative of characteristics relevant to predictions associated with the adverse event, providing a second set of training data, determining, by a trigger module, a trigger decision based on a set of signals at least partially determined from the second set of training data, the trigger decision indicating whether the ML model is to be one of updated and retrained based on the second set of training data, and selectively executing one of updating and retraining of the ML model using at least a portion of the second set of training data in response to the trigger decision.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for developing electrical grid mapping. One of the methods includes obtaining a computer model of an electric power grid; generating a network graph representation of the computer model, wherein nodes of the network graph represent grid assets of the computer model and edges of the network graph represent wires connecting the grid assets; generating an initial prediction of links between nodes in the network graph by adding at least one edge to the network graph to obtain an over-connected graph; applying the over-connected network graph as input to a machine learning model to obtain an annotated network graph, the machine learning model configured to identify edges as positive links and negative links, and apply annotations to the edges indicating whether each edge is a positive or negative link; and updating the model based on the annotated network graph.

Abstract: Disclosed implementations relate to using mutual constraint satisfaction to sample from different stochastic processes and identify coherent inferences across domains. In some implementations, a first domain representation of a semantic concept may be used to conditionally sample a first set of candidate second domain representations of the semantic concept from a first stochastic process. Based on second domain representation(s) of the first set, candidate third domain representations of the semantic concept may be conditionally sampled from a second stochastic process. Based on candidate third domain representation(s), a second set of candidate second domain representations of the semantic concept may be conditionally sampled from a third stochastic process. Pairs of candidate second domain representations sampled across the first and second sets may be evaluated. Based on the evaluation, second domain representation(s) of the semantic concept are selected, e.g., as input for a downstream computer process.

Abstract: A method for electrical grid monitoring using aggregated smart meter data includes receiving, from an electric grid monitoring device, electrical data aggregated from a plurality of electric meters each connected to a secondary side of at least one electrical transformer within a local geographic region. Each electric meter is configured to generate measurement data based on measurements taken at a respective electrical load. The electrical data indicates one or more characteristics of a secondary electrical distribution system connected to the at least one electrical transformer. The method includes determining, based on the electrical data aggregated from the plurality of electric meters, one or more characteristics of the primary electrical distribution system connected to the electrical transformer.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a storage device, for identifying characteristics of electric grid assets are disclosed. A method includes obtaining a plurality of images, each image depicting at least one utility pole of an electric grid; detecting, in each of the plurality of images, keypoints of the at least one utility pole depicted in the image; determining, using the keypoints from at least two images of a particular utility pole, at least one measurement of the particular utility pole; determining, using the measurement of the particular utility pole, an electrical characteristic of an asset supported by the particular utility pole; and providing the electrical characteristic as an output. The two or more images include images of the particular utility pole captured from multiple different camera perspectives. The asset can include a capacitor, a transformer, a switch, a power line.

Abstract: Implementations are described herein for aggregating information responsive to a query from multiple different data feed services using machine learning. In various implementations, NLP may be performed on a natural language input comprising a query for information to generate a data feed-agnostic aggregator embedding (FAAE). A plurality of data feed services may be selected, each having its own data feed service action space that includes actions that are performable to access data via the data feed service. The FAAE may be processed based on domain-specific machine learning models corresponding to the selected data feed services. Each domain-specific machine learning model may translate between a respective data feed service action space and a data feed-agnostic semantic embedding space. Using these models, action(s) may be selected from the data feed service action spaces and performed to aggregate, from the plurality of data feed services, data that is responsive to the query.

Abstract: A 2×2 photonic splitter includes two mode converting Y-junctions. A first stage mode converting Y-junction includes input branch ports adapted to receive an input optical signal propagating in a fundamental spatial mode at either of the input branch ports, a first trunk port, and a first mode converting region. The first mode converting region is adapted to convert at least a first power portion of the fundamental spatial mode of the input optical signal when received via at least one of the input branch ports to a higher order spatial mode at the first trunk port. The second stage mode converting Y-junction includes output branch ports adapted to emit output optical signals having the fundamental spatial mode, a second trunk port, and a second mode converting region optically coupling the output branch ports to the second trunk port. A connected trunk section photonically links the trunk ports.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for a multi-modal artificial intelligence platform for building construction. The multi-model artificial intelligence platform includes various engines that perform various computer-implemented methods. The various engines include a site selector artificial intelligence (AI) engine, a geospatial database, a site compliance analyzer AI engine, a masterplan generator AI engine, a compliance analyzer AI engine, an aesthetic generator AI engine, a schematic generator AI engine, a construction plan generator AI engine, a project timeline generator AI engine, a compliance application generator AI engine, and a financial model generator AI engine. Respective AI engines collaboratively cooperate for end-to-end AI-driven building construction design and development.

Abstract: Methods, systems, and apparatus, including medium-encoded computer program products, for configuring location-specific electrical load models while preserving privacy. A first machine learning model configured to predict electrical load curves of an electrical utility grid can be obtained from a server. Load values associated with a particular region of the electrical utility grid can be obtained. The load values can be applied as calibration input to the first machine learning model to produce first adjustment parameters for the first machine learning model. The first adjustment parameters can be provided to the server.

Abstract: Methods, computer systems, and apparatus, including computer programs encoded on computer storage media, for training a classification neural network. The system generates, from a set of object-specific data, one or more meta-learning datasets for one or more respective initial training tasks. The system determines values for a set of meta parameters by performing meta-learning with a classification neural network on the one or more meta-learning datasets. The system obtains a set of labeled training examples for a characteristic-detection task. The system determines based at least on one of the values for the set of meta parameters and using the set of labeled training examples, target values for the network parameters for the classification neural network to perform the characteristic-detection task.

Abstract: A method for generating spectroscopic data includes inputting, by a computing device, a text string comprising a structural representation of a material to an encoder of a natural language processing (NLP) model implemented with a deep neural network. The method includes generating, using the encoder of the NLP model, an encoded representation of the text string. The text string may include latent chemical bond information of the material. The method includes mapping, by the computing device, the encoded representation including the latent chemical bond information to a spectrum array, the spectrum array including predicted spectroscopic data of the material. The method also includes outputting, by the computing device, the spectrum array.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for characterization of aggregate particles. A method includes obtaining, from a set of low fidelity sensors, first sensor data of a first portion of particles; obtaining, from a set of high fidelity sensors, second sensor data of the first portion of particles, the second sensor data comprising a higher fidelity representation of characteristics of the first portion of particles than the first sensor data; training a characterization model using the first sensor data and the second sensor data, the training comprising: providing, as training data to the characterization model, the second sensor data; and processing the second sensor data with the characterization model to correlate the first sensor data with the second sensor data. The first sensor data can indicate shape characteristics of each particle; and the second sensor data indicates a surface area of each particle.

Abstract: Methods, systems, and apparatus for receiving a request for a risk assessment for a parcel, receiving a set of images for the parcel, the set of images including two or more images, each image having an image scale and an image resolution that is different from other images in the set of images, providing a first-level feature embedding and a second-level feature embedding, the first-level feature embedding being provided by processing a first-level image through a first-level machine learning (ML) model, and the second-level feature embedding being provided by processing a second-level image through a second-level ML model, determining a risk assessment at least partially by processing each of the first-level feature embedding and a second-level feature embedding through a fusion network, and providing a representation of the risk assessment for display.

Abstract: Methods and systems for using multiple hyperspectral cameras sensitive to different wavelengths to predict characteristics of objects for further processing, including recycling, are described. The multiple hyperspectral images can be used to predict higher resolution spectra by using a trained machine learning model. The higher resolution spectra may be more easily analyzed to sort plastics into a recyclability category. The hyperspectral images may also be used to identify and analyze dark or black plastics, which are challenging for SWIR, MWIR, and other wavelengths. The machine learning model may also predict the base polymers and contaminants of plastic objects for recycling. The hyperspectral images may be used to predict recyclability and other characteristics using a trained machine learning model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for improved image segmentation using hyperspectral imaging. In some implementations, a system obtains image data of a hyperspectral image, the image data comprising image data for each of multiple wavelength bands. The system accesses stored segmentation profile data for a particular object type that indicates a predetermined subset of the wavelength bands designated for segmenting different region types for images of an object of the particular object type. The system segments the image data into multiple regions using the predetermined subset of the wavelength bands specified in the stored segmentation profile data to segment the different region types. The system provides output data indicating the multiple regions and the respective region types of the multiple regions.

Abstract: A polarization beam splitter includes an input port, first and second output ports, and a polarization splitting region coupled between the input port and the first and second output ports. The input port is adapted to receive guided optical signals that are polarization multiplexed, including a transverse electric (TE) optical signal and a transverse magnetic (TM) optical signal. The polarization splitting region includes a pattern of at least two materials having different refractive indexes. The pattern is shaped to demultiplex the TE and TM optical signals by directing a first power majority of the TE optical signal received at the input port to the second output port via asymmetrical power splitting while directing a second power majority of the TM optical signal received at the input port to the first output port via multipath interferometry.

Abstract: Computer-implemented methods may include accessing a multi-dimensional embedding space that supports relating embeddings of molecules to predicted values of a given property of the molecules. The method may also include identifying one or more points of interest within the embedding space based on the predicted values. Each of the one or more points of interest may include a set of coordinate values within the multi-dimensional embedding space and may be associated with a corresponding predicted value of the given property. The method may further include generating, for each of the one or more points of interest, a structural representation of a molecule by transforming the set of coordinate values included in the point of interest using a decoder network. The method may include outputting a result that identifies, for each of the one or more points of interest, the structural representation of the molecule corresponding to the point of interest.