We show how randomly scrambling the output classes of various fractions of the training data may be used to improve predictive accuracy of a classification algorithm. We present a method for calculating the "noise sensitivity signature" of a learning algorithm which is based on scrambling the output classes. This signature can be used to indicate a good match between the complexity of the classifier and the complexity of the data. Use of noise sensitivity signatures is distinctly different from other schemes to avoid over(cid:173) training, such as cross-validation, which uses only part of the train(cid:173) ing data, or various penalty functions, which are not data-adaptive. Noise sensitivity signature methods use all of the training data and are manifestly data-adaptive and non-parametric.

An artificial-intelligence art project has been criticised for using racist and sexist tags to classify its users. When they share a selfie with ImageNet Roulette, the web app matches it to the ones it most closely resembles from an enormous library of profile photos. It then reveals the most popular tag, assigned to the matching pictures by human workers using data set WordNet. These include racial slurs, "first offender", "rape suspect", "spree killer", "newsreader", and "Batman". Those responsible for assigning the tags to the library pictures were recruited via a service offered by Amazon, called Mechanical Turk, which pays workers around the world pennies to perform small, monotonous tasks.

We show how randomly scrambling the output classes of various fractions of the training data may be used to improve predictive accuracy of a classification algorithm. We present a method for calculating the "noise sensitivity signature" of a learning algorithm which is based on scrambling the output classes. This signature can be used to indicate a good match between the complexity of the classifier and the complexity of the data. Use of noise sensitivity signatures is distinctly different from other schemes to avoid overtraining, such as cross-validation, which uses only part of the training data, or various penalty functions, which are not data-adaptive. Noise sensitivity signature methods use all of the training data and are manifestly data-adaptive and nonparametric. They are well suited for situations with limited training data. 1 INTRODUCTION A major problem of pattern recognition and classification algorithms that learn from a training set of examples is to select the complexity of the model to be trained. How is it possible to avoid an overparameterized algorithm from "memorizing" the training data?

We show how randomly scrambling the output classes of various fractions of the training data may be used to improve predictive accuracy of a classification algorithm. We present a method for calculating the "noise sensitivity signature" of a learning algorithm which is based on scrambling the output classes. This signature can be used to indicate a good match between the complexity of the classifier and the complexity of the data. Use of noise sensitivity signatures is distinctly different from other schemes to avoid overtraining, such as cross-validation, which uses only part of the training data, or various penalty functions, which are not data-adaptive. Noise sensitivity signature methods use all of the training data and are manifestly data-adaptive and nonparametric. They are well suited for situations with limited training data. 1 INTRODUCTION A major problem of pattern recognition and classification algorithms that learn from a training set of examples is to select the complexity of the model to be trained. How is it possible to avoid an overparameterized algorithm from "memorizing" the training data?

We show how randomly scrambling the output classes of various fractions of the training data may be used to improve predictive accuracy of a classification algorithm. We present a method for calculating the "noise sensitivity signature" of a learning algorithm which is based on scrambling the output classes. This signature can be used to indicate a good match between the complexity of the classifier and the complexity of the data. Use of noise sensitivity signatures is distinctly different from other schemes to avoid overtraining, suchas cross-validation, which uses only part of the training data, or various penalty functions, which are not data-adaptive. Noise sensitivity signature methods use all of the training data and are manifestly data-adaptive and nonparametric.