Adaptive filtering and classification for automated tuning of quantum dots into the single electron regime

One of the key challenges towards the scalability of spin qubits lies in the control and calibration of quantum dots. The gate voltages required to reach a desired charge state are different for each qubit thus requiring a lengthy and complex characterization process. This work presents an algorithm for automated tuning of single quantum dots into the single electron regime that is independent of the measurement system used. Although it is reliant on a small transition line width, the fast computation time and low amount of data required to reach the desired regime makes it an efficient tuning tool for quantum dots. The algorithm navigates the gate voltage space by performing sparse measurements of stability diagrams. Each measurement is filtered using an adaptive thresholding technique based on an exponentially weighted moving average. The filtered data is then classified using an established neural network approach that we upgraded with a line orientation detection module to help with the decision making. Combining this method with Keysight's fast measurement platform, reaching the single electron regime can be done in a matter of seconds while measuring only a small fraction of the data usually necessary when using full stability diagram measurements.