Adiabatic quantum learning

Quantum machine learning has attracted considerable interest due to its potential to improve certain learning tasks. In conventional quantum machine learning, the output is the expectation value of a preselected observable, and the projective measurement forces a quantum circuit to run many times to obtain the output with reasonable precision. In this work, we propose a protocol to utilize the adiabatic quantum evolution to execute quantum learning tasks, in which the output is obtained by the adiabatic weak measurement rather than the projective measurement. In comparison to previous protocols, we use only a single-shot measurement and therefore avoid the measurement repetition in the previous protocols. Moreover, our protocol allows us to extract the expectation values of multiple observables without disrupting the concerned quantum states.