Problem
Quantum State Tomography experiments rely on multiple measurements to reconstruct an unknown quantum state. This disregards the fact that some measurements may be less informative than others. This introduces a factor of inefficiency that scales with the number of measurements needed.
Approach
Active-learning can addresses this inefficiency by selecting the best basis to conduct a measurement in, rather than measuring in each basis equally.
Outcome
Active-learning was able to reach peak fidelity faster for small-shot experiments than traditional measurement. In the long-shot, active-learning was able to acheive a ~0.2% higher fidelity than traditional measurement. In our findings, we concluded that this was a very useful method in aiding Quantum State Tomography experiments.