Towards interpretable quantum machine learning via single-photon quantum walks
This work addresses interpretability for quantum reinforcement learning agents, representing an incremental step by adapting an existing classical model to a quantum framework.
The authors tackled the lack of interpretability in quantum machine learning by developing a variational method to quantize projective simulation, a reinforcement learning model, using single-photon quantum walks. They demonstrated that the quantized model can exploit quantum interference to achieve capabilities beyond its classical counterpart in a transfer learning example.
Variational quantum algorithms represent a promising approach to quantum machine learning where classical neural networks are replaced by parametrized quantum circuits. However, both approaches suffer from a clear limitation, that is a lack of interpretability. Here, we present a variational method to quantize projective simulation (PS), a reinforcement learning model aimed at interpretable artificial intelligence. Decision making in PS is modeled as a random walk on a graph describing the agent's memory. To implement the quantized model, we consider quantum walks of single photons in a lattice of tunable Mach-Zehnder interferometers trained via variational algorithms. Using an example from transfer learning, we show that the quantized PS model can exploit quantum interference to acquire capabilities beyond those of its classical counterpart. Finally, we discuss the role of quantum interference for training and tracing the decision making process, paving the way for realizations of interpretable quantum learning agents.