Training-Efficient Density Quantum Machine Learning
A density quantum neural network is a mixture of trainable circuits. Cut a circuit into commuting pieces and the mixture trains at backpropagation cost. Stack copies and it grows without a deeper circuit.
Quantum Deep Hedging
Quantum neural networks with orthogonal and compound layers act as the policy and the critic in Deep Hedging. A distributional actor-critic gives the best policies, and the models run on a trapped-ion processor with up to 16 qubits.