Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for determining neural network architectures. One of the methods includes selecting a candidate architecture; selecting a neural network block from the set of neural network blocks; determining whether to (i) add the selected neural network block as a new neural network block in the candidate architecture or (ii) replace one of the neural network blocks in the selected candidate architecture with the selected neural network block; based on the determining, generating a mutated architecture; training a neural network having the mutated architecture on the training data; determining a performance measure for the trained neural network that measures the performance of the trained neural network on the particular machine learning task; and adding, to the maintained data, data specifying the mutated architecture and data associating the mutated architecture with the determined performance measure.

Abstract: A measurement of an effect of a topic on an aggregate of numerical information related to a set of evaluations of a specific product can be produced. A subset of the set of the evaluations can be determined. The subset can be defined by inclusion of textual information about the topic. The specific product can be a good, a service, an application, the like, or any combination thereof. An aggregate of the numerical information related to the subset can be determined. Based on the aggregate of the numerical information related to the subset, the measurement of the effect of the topic on the aggregate of the numerical information related to the set can be calculated. The measurement can be included in a file to be transmitted to a computer system to be used to control operations performed by the computer system to produce a modification to the specific product.

Abstract: A measurement of an effect of a topic on an aggregate of numerical information related to a set of evaluations of a specific product can be produced. A subset of the set of the evaluations can be determined. The subset can be defined by inclusion of textual information about the topic. The specific product can be a good, a service, an application, the like, or any combination thereof. An aggregate of the numerical information related to the subset can be determined. Based on the aggregate of the numerical information related to the subset, the measurement of the effect of the topic on the aggregate of the numerical information related to the set can be calculated. The measurement can be included in a file to be transmitted to a computer system to be used to control operations performed by the computer system to produce a modification to the specific product.