Title: Some aspects of noise in binary classification with quantum circuits

Abstract: We formally study the effects of a restricted single-qubit noise model inspired by real quantum hardware, and corruption in quantum training data, on the performance of binary classification using quantum circuits. We find that, under the assumptions made in our noise model, that the measurement of a qubit is affected only by the noises on that qubit even in the presence of entanglement. Furthermore, when fitting a binary classifier using a quantum dataset for training, we show that noise in the data can work as a regularizer, implying potential benefits from the noise in certain cases for machine learning problems.