Amplitude embedding lets a variational quantum model read many classical features with few qubits, but the model output is then only a quadratic form in the normalized input. Data re-uploading, which...
arXiv:2604. 06135v2 Announce Type: replace-cross Abstract: Efficient data loading remains a bottleneck for near-term quantum machine learning.
By Basil Kyriacou, Viktoria Patapovich, Maniraman Periyasamy, Alexey Melnikov
arXiv:2504. 05336v4 Announce Type: replace-cross Abstract: A recurring weakness in quantum machine learning (QML) is that reported ``quantum advantages'' are seldom tested against a \emph{capacity-matched} classical control, leaving it unclear whether a gain comes from the quantum substrate or from the architectural change that accompanies it.
By Chi-Sheng Chen, En-Jui Kuo
arXiv:2608. 14691v1 Announce Type: new Abstract: Sequence models are conventionally distinguished by their backbone, the mechanism that routes information across positions, such as attention or recurrence.
By Ahmed Nebli, Hadi Saadatdoorabi, Christopher Keibel, Kevin Yam
The paper compares single‑qubit re‑uploading (serial) variational quantum circuits with multi‑qubit circuits that use the same encoding budget for one‑dimensional inputs. It shows that although a serial circuit can represent any function a multi‑qubit circuit can, its first‑order training is more costly because all parameters move the state along a single phase‑locking direction, limiting the number of effective tangent directions. Experiments demonstrate that for larger encoding budgets the serial circuit converges far less often than the multi‑qubit control, and its held‑out error can be up to 1.5 times higher under Adam optimization.
By Michael Poppel, David Bucher, Maximilian Zorn, Markus Baumann, Sebastian W\"olckert, Claudia Linnhoff-Popien, Philipp Altmann, Jonas Stein
arXiv:2602. 06699v2 Announce Type: replace-cross Abstract: We propose a variational quantum implementation of self-attention (QSA)-the core operation in transformers and large language models-which predicts future elements of a sequence by forming overlap-weighted combinations of past data.
By Alessio Pecilli, Matteo Rosati