Abstract: Examples are disclosed that relate to a generative model for generating inorganic material candidates, such as crystalline structures. One example provides a method, comprising training an unconditional generative model using a dataset of stable periodic material structures, the unconditional generative model comprising a diffusion model. The training comprises learning the diffusion model to iteratively noise the stable periodic material structures of the dataset towards a random periodic structure by noising atom types of atoms in the periodic material structure, noising fractional coordinates of the atoms in the periodic material structure, and noising a lattice of the periodic material structure. The method further comprises using the trained unconditional generative model to generate a material structure by iteratively denoising an initial structure sampled from a random distribution.

Abstract: A computation graph of a machine learning model is accessed from memory and a constraint solver is used to compute a partition of the computation graph into ordered stages of an execution pipeline. In use, when inference or training of the machine learning model takes place by executing the pipeline, execution cost of the stages are balanced according to the computed partition.

Abstract: A sensor element for detecting a target gas to be measured in a measurement-object gas includes: a base part in an elongated plate shape, including an oxygen-ion-conductive solid electrolyte layer; a measurement-object gas flow part formed from one end part in a longitudinal direction of the base part; an inner main pump electrode disposed on an inner surface of the measurement-object gas flow part; a porous coating layer covering at least an electrode end part of the inner main pump electrode closer to the one end part in the longitudinal direction of the base part; and a measurement electrode disposed at a position farther from the one end part in the longitudinal direction of the base part than the inner main pump electrode on the inner surface of the measurement-object gas flow part.

Abstract: A neural network training apparatus is described which has a network of worker nodes each having a memory storing a subgraph of a neural network to be trained. The apparatus has a control node connected to the network of worker nodes. The control node is configured to send training data instances into the network to trigger parallelized message passing operations which implement a training algorithm which trains the neural network. At least some of the message passing operations asynchronously update parameters of individual subgraphs of the neural network at the individual worker nodes.

Abstract: A sensor element includes: a base part in an elongated plate shape; a measurement-object gas flow part formed from one end part in a longitudinal direction of the base part; an inner main pump electrode disposed on an inner surface of the measurement-object gas flow part; an outer pump electrode disposed in correspondence with the inner main pump electrode; and a measurement electrode disposed at a position farther from the one end part than the inner main pump electrode on the inner surface of the measurement-object gas flow part. The outer pump electrode is disposed at a position farther from the one end part in the longitudinal direction of the base part than one electrode end of the inner main pump electrode closer to the one end part in the longitudinal direction of the base part, with respect to the longitudinal direction of the base part.

Abstract: A neural network training apparatus is described which has a network of worker nodes each having a memory storing a subgraph of a neural network to be trained. The apparatus has a control node connected to the network of worker nodes. The control node is configured to send training data instances into the network to trigger parallelized message passing operations which implement a training algorithm which trains the neural network. At least some of the message passing operations asynchronously update parameters of individual subgraphs of the neural network at the individual worker nodes.

Abstract: A computation graph of a machine learning model is accessed from memory and a constraint solver is used to compute a partition of the computation graph into ordered stages of an execution pipeline. In use, when inference or training of the machine learning model takes place by executing the pipeline, execution cost of the stages are balanced according to the computed partition.

Abstract: A computerized method for forecasting a future conformation of a molecular system based on a current conformation of the molecular system comprises (a) receiving the current conformation in a trained machine-learning model that has been previously trained to map a plurality of conformations received to a corresponding plurality of conformations proposed; (b) mapping the current conformation to a proposed conformation via the trained machine-learning model, wherein the proposed conformation is appended to a Markov chain; and (c) returning the proposed conformation as the future conformation.

Abstract: A neural network training apparatus is described which has a network of worker nodes each having a memory storing a subgraph of a neural network to be trained. The apparatus has a control node connected to the network of worker nodes. The control node is configured to send training data instances into the network to trigger parallelized message passing operations which implement a training algorithm which trains the neural network. At least some of the message passing operations asynchronously update parameters of individual subgraphs of the neural network at the individual worker nodes.

Abstract: A neural network training apparatus is described which has a network of worker nodes each having a memory storing a subgraph of a neural network to be trained. The apparatus has a control node connected to the network of worker nodes. The control node is configured to send training data instances into the network to trigger parallelized message passing operations which implement a training algorithm which trains the neural network. At least some of the message passing operations asynchronously update parameters of individual subgraphs of the neural network at the individual worker nodes.

Abstract: A computation graph of a machine learning model is accessed from memory and a constraint solver is used to compute a partition of the computation graph into ordered stages of an execution pipeline. In use, when inference or training of the machine learning model takes place by executing the pipeline, execution cost of the stages are balanced according to the computed partition.

Abstract: Disclosed in some examples are methods, systems, machine-readable media, and devices which operate a neural network defined by user code. A method includes identifying, operations from user code that are integral in operating the neural network, combining a subset of the identified operations into a single processing sequence to be transmitted to an array of hardware processors, performing operations that are not integral in operation of the neural network in a separate thread of execution from the operations that are integral in operating the neural network; and mapping results to the combined operations that were included in the single processing sequence.

Abstract: A data compression apparatus is described which has an encoder configured to receive an input data item and to compress the data item into an encoding comprising a plurality of numerical values. The numerical values are grouped at least according to whether they relate to content of the input data item or style of the input data item. The encoder has been trained using a plurality of groups of training data items grouped according to the content and where training data items within individual ones of the groups vary with respect to the style. The encoder has been trained using a training objective which takes into account the groups.

Abstract: A data compression apparatus is described which has an encoder configured to receive an input data item and to compress the data item into an encoding comprising a plurality of numerical values. The numerical values are grouped at least according to whether they relate to content of the input data item or style of the input data item. The encoder has been trained using a plurality of groups of training data items grouped according to the content and where training data items within individual ones of the groups vary with respect to the style. The encoder has been trained using a training objective which takes into account the groups.

Abstract: A data compression apparatus is described which has an encoder configured to receive an input data item and to compress the data item into an encoding comprising a plurality of numerical values. The numerical values are grouped at least according to whether they relate to content of the input data item or style of the input data item. The encoder has been trained using a plurality of groups of training data items grouped according to the content and where training data items within individual ones of the groups vary with respect to the style. The encoder has been trained using a training objective which takes into account the groups.

Abstract: Disclosed in some examples are methods, systems, machine-readable media, and devices which operate a neural network defined by user code. A method includes identifying, operations from user code that are integral in operating the neural network, combining a subset of the identified operations into a single processing sequence to be transmitted to an array of hardware processors, performing operations that are not integral in operation of the neural network in a separate thread of execution from the operations that are integral in operating the neural network; and mapping results to the combined operations that were included in the single processing sequence.

Abstract: A neural network training apparatus is described which has a network of worker nodes each having a memory storing a subgraph of a neural network to be trained. The apparatus has a control node connected to the network of worker nodes. The control node is configured to send training data instances into the network to trigger parallelized message passing operations which implement a training algorithm which trains the neural network. At least some of the message passing operations asynchronously update parameters of individual subgraphs of the neural network at the individual worker nodes.

Abstract: A data compression apparatus is described which has an encoder configured to receive an input data item and to compress the data item into an encoding comprising a plurality of numerical values. The numerical values are grouped at least according to whether they relate to content of the input data item or style of the input data item. The encoder has been trained using a plurality of groups of training data items grouped according to the content and where training data items within individual ones of the groups vary with respect to the style. The encoder has been trained using a training objective which takes into account the groups.

Abstract: A computation node of a neural network training system is described. The node has a memory storing a plurality of gradients of a loss function of the neural network and an encoder. The encoder encodes the plurality of gradients by setting individual ones of the gradients either to zero or to a quantization level according to a probability related to at least the magnitude of the individual gradient. The node has a processor which sends the encoded plurality of gradients to one or more other computation nodes of the neural network training system over a communications network.