BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Causal Discovery with Quantum Machine Learning
DTSTART;VALUE=DATE-TIME:20260910T101000Z
DTEND;VALUE=DATE-TIME:20260910T104000Z
DTSTAMP;VALUE=DATE-TIME:20260922T191358Z
UID:indico-contribution-2067@events.ncbj.gov.pl
DESCRIPTION:Speakers: Piotr Gawron (CAMK PAN)\nDetecting statistical depen
 dence between stochastic processes is a core primitive of causal discovery
  for dynamical systems. State-of-the-art tests compare whole trajectories 
 with the signature kernel\, which propagates a static point-similarity ker
 nel through a Goursat PDE and feeds the resulting Gram matrices into kerne
 l independence tests (HSIC\, SDCIT). We investigate replacing the classica
 l RBF static kernel with a quantum embedding kernel: a data-reuploading ci
 rcuit maps each time-augmented path point to a quantum state\, and similar
 ity is the state fidelity\, read out with the adjoint-circuit trick. Since
  only the static kernel is swapped\, performance differences are attributa
 ble to the kernel itself. The circuit angles and an input scale are traine
 d to maximize a standardized-HSIC proxy of test power using SPSA with Adam
  and held-out validation\, avoiding differentiation through the PDE solver
 . In simulations on linear SDEs\, the trained quantum kernel improves HSIC
  test power over both the classical and the untrained quantum kernel at sm
 all sample sizes\, while preserving type-I error control\, and supports PC
 -style recovery of multivariate causal graphs with conditional tests. We f
 urther study register widths of 4-12 qubits\, connecting fidelity concentr
 ation to trainability\, and note that the ansatz maps natively onto IBM He
 ron hardware (CRZ as native RZZ plus virtual RZ\; 12-qubit heavy-hex loops
 ).\n\nhttps://events.ncbj.gov.pl/event/468/contributions/2067/
LOCATION:
URL:https://events.ncbj.gov.pl/event/468/contributions/2067/
END:VEVENT
END:VCALENDAR