Quantum Technology Initiative Journal Club

Thursday 24 Oct 2024, 16:00 → 17:00 Europe/Zurich

513/R-070 - Openlab Space (CERN)

Michele Grossi (CERN)

Weekly Journal Club meetings organised in the framework of the CERN Quantum Technology Initiative (QTI) to present and discuss scientific papers in the field of quantum science and technology. The goal is to help researchers keep track of current findings and walk away with ideas for their own research. Some previous knowledge of quantum physics would be helpful, but is not required to follow the talks.

To propose a paper for discussion, contact: michele.grossi@cern.ch

16:00 → 17:00 CERN QTI Journal CLUB: TITLE

Convener: Dr Michele Grossi (CERN)

16:00 José Ramón Pareja Monturiol (Univ. Madrid)

TITLE: Tensorizing high-dimensional densities

LINK:
NA

ABSTRACT:
Tensor networks are compressed and efficient representations of high-dimensional tensors that have found successful applications in physics, mathematics and machine learning. A significant challenge is determining the smaller tensors that constitute the network, when these cannot be derived analytically. In such cases, two approaches are typically employed: optimization techniques or direct factorization methods.In our work, we propose an adaptation of an algorithm of the second type (Tensor Train Recursive Sketching), that is applicable when a high-dimensional density is provided along with a small set of samples. To evaluate the performance of the algorithm, we investigate two different applications: machine learning and condensed matter physics. In the former application, we apply the decomposition to derive a Tensor Train model from a pre-trained neural network binary classifier. We demonstrate that this translation from a black-box model, the neural network, to a tensor network can be advantageous in terms of privacy, interpretability and efficiency. For the latter application in condensed matter physics, we show that TT-RSS enables the reconstruction of the AKLT state from black-box observations of the same state. Although this serves as a sanity check, it paves the way for more intriguing applications, such as finding TN representations of Quantum Neural States to enhance interpretability.

Speaker: JJosé Ramón Pareja Monturiol (Univ. Madrid)

recording