Abstract: Neural network-based categorization can be improved by incorporating graph neural networks that operate on a graph representing the taxonomy of the categories into which a given input is to be categorized by the neural network based-categorization. The output of a graph neural network, operating on a graph representing the taxonomy of categories, can be combined with the output of a neural network operating upon the input to be categorized, such as through an interaction of multidimensional output data, such as a dot product of output vectors. In such a manner, information conveying the explicit relationships between categories, as defined by the taxonomy, can be incorporated into the categorization. To recapture information, incorporate new information, or reemphasize information a second neural network can also operate upon the input to be categorized, with the output of such a second neural network being merged with the output of the interaction.

Abstract: In some aspects, a method of carbonation control can include monitoring, via a first sensor, an environmental condition in a contactor unit, the contactor unit including a plurality of carbonation vessels including carbonation medium, monitoring, via a second sensor, an environmental condition at a location outside of the contactor unit, predicting, based on the measured environmental condition in the contactor unit and the measured environmental condition at the location outside of the contactor unit, a local humidity in a carbonation vessel from the plurality of carbonation vessels, a water content of the carbonation medium in the carbonation vessel from the plurality of carbonation vessels, and a carbonation extent of the carbonation medium in the carbonation vessel from the plurality of carbonation vessels for a carbonation forecasting period, and executing an action on the plurality of carbonation vessels based on the predicted water content and predicted carbonation extent.