Systems and methods for expanding a multi-relational data structure tunable for generating a non-linear dataset from a time-dependent query. The systems include a processor and a memory. The memory may store processor-executable instructions that, when executed, configure the processor to: receive the query of the multi-relational data structure, wherein the query includes at least one entity node at a queried time relative to the time data; obtain, based on the query, a temporal representation vector based on a diachronic embedding of the multi-relational data structure, the diachronic embedding based on a combination of a first sub-function associated with a temporal feature and a second sub-function associated with a persistent feature; determine, from the temporal representation vector, at least one time-varied score corresponding to the queried time; and generate a response dataset based on the at least one time-varied score determined from the temporal representation vector.

The evolutionary feature selection algorithm is combined with model evaluation during training to learn feature subsets that maximize speech/non-speech distribution distances. The technique enables ensembling of low-cost models over similar features subspaces increases classification accuracy and has similar computational complexity in practice. Prior to training the models, feature analysis is conducted via an evolutionary feature selection algorithm which measures fitness for each feature subset in the population by its k-fold cross validation score. PCA and LDA based eigen-features are computed for each subset and fitted with a Gaussian Mixture Model from which combinations of feature subsets with Maximum Mean Discrepancy scores are obtained. During inference, the resulting features are extracted from the input signal and given as input to the trained neural networks.

The evolutionary feature selection algorithm is combined with model evaluation during training to learn feature subsets that maximize speech/non-speech distribution distances. The technique enables ensembling of low-cost models over similar features subspaces increases classification accuracy and has similar computational complexity in practice. Prior to training the models, feature analysis is conducted via an evolutionary feature selection algorithm which measures fitness for each feature subset in the population by its k-fold cross validation score. PCA and LDA based eigen-features are computed for each subset and fitted with a Gaussian Mixture Model from which combinations of feature subsets with Maximum Mean Discrepancy scores are obtained. During inference, the resulting features are extracted from the input signal and given as input to the trained neural networks.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for implementing brain emulation neural networks using compressed matrix representations. One of the methods includes obtaining a network input; and processing the network input using a neural network to generate a network output, comprising: processing the network input using an input subnetwork of the neural network to generate an embedding of the network input; and processing the embedding of the network input using a brain emulation subnetwork of the neural network, wherein the brain emulation subnetwork has a brain emulation neural network architecture that represents synaptic connectivity between a plurality of biological neurons in a brain of a biological organism, the processing comprising: obtaining a compressed matrix representation of a sparse matrix of brain emulation parameters; and applying the compressed matrix representation to the embedding of the network input to generate a brain emulation subnetwork output.

Technology for inspection for detecting a defect of a printed matter using machine logic informed by machine learning. Some embodiments of the present invention may include one, or more, of the following features: (i) generates defect datasets; (ii) generates defect libraries; (iii) uses the generated defect libraries for deep learning training; and (iv) uses machine learning to detect defects using computer code (for example, a *.jpg format file) corresponding to an image of a piece of printed matter instead of using a visual image (that is, an image of the type that is created when a person takes a picture using a traditional film camera).

An evolutionary AutoML framework called LEAF optimizes hyperparameters, network architectures and the size of the network. LEAF makes use of both evolutionary algorithms (EAs) and distributed computing frameworks. A multiobjective evolutionary algorithm is used to maximize the performance and minimize the complexity of the evolved networks simultaneously by calculating the Pareto front given a group of individuals that have been evaluated for multiple objectives.

Respective labels indicative of compression-related quality degradation for a set of media object tuples which meet a divergence criterion are obtained; each tuple comprises a reference media object and a pair of corresponding compressed media object versions. Pairs of training records for a machine learning model are generated using the labeled media object tuples and multiple perceptual quality algorithms, with each training record comprising respective perceived quality degradation scores generated by each of the multiple algorithms for a given compressed media object of a tuple. A machine learning model is trained, using the record pairs, to predict quality degradation scores for compressed media objects.

Training an encoder is provided. The method comprises inputting a current state of a number of aircraft into a recurrent layer of a neural network, wherein the current state comprises a reduced state in which a value of a specified parameter is missing. An action applied to the aircraft is input into the recurrent layer concurrently with the current state. The recurrent layer learns a value for the parameter missing from current state, and the output of the recurrent layer is input into a number of fully connected hidden layers. The hidden layers, according to the current state, learned value, and current action, determine a residual output that comprises an incremental difference in the state of the aircraft resulting from the current action.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training a segmentation neural network. In one aspect, a method comprises: obtaining data defining: (i) an image, and (ii) a respective class of each pixel in the image from a set of possible classes; determining a target segmentation of the image that comprises one or more target contrastive channels, wherein each target contrastive channel corresponds to a respective pair of classes including a respective first class and a respective second class from the set of possible classes; and training the segmentation neural network to process the image to generate a predicted segmentation that matches the target segmentation.

In one aspect, there is provided a method performed by one or more data processing apparatus for selecting actions to be performed by an agent interacting with an environment, the method including, at each of multiple time steps, receiving an observation characterizing a current state of the environment at the time step, providing an input including the observation to an action selection neural network having a brain emulation sub-network with an architecture that is based on synaptic connectivity between biological neurons in a brain of a biological organism, processing the input including the observation characterizing the current state of the environment at the time step using the action selection neural network having the brain emulation sub-network to generate an action selection output, and selecting an action to be performed by the agent at the time step based on the action selection output.