BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:PILA: Physics-Informed Low Rank Augmentation for Interpretable Ear
 th Observation - Yihang She\, University of Cambridge
DTSTART:20260313T130000Z
DTEND:20260313T140000Z
UID:TALK241933@talks.cam.ac.uk
CONTACT:114742
DESCRIPTION:*Abstract*\n\nPhysically meaningful representations are essent
 ial for Earth Observation (EO)\, yet existing physical models are often si
 mplified and incomplete. This leads to discrepancies between simulation an
 d observations that hinder reliable forward model inversion. Common approa
 ches to EO inversion either ignored this incompleteness or relied on case-
 specific pre-processing. More recent methods use physics-informed autoenco
 ders but depend on auxiliary variables that are difficult to interpret and
  multiple regularizers that are difficult to balance. We propose Physics-I
 nformed Low-Rank Augmentation (PILA)\, a framework that augments incomplet
 e physical models using a learnable low-rank residual to improve flexibili
 ty\, while remaining close to the governing physics. \nWe evaluate PILA on
  two EO inverse problems involving diverse physical processes: forest radi
 ative transfer inversion from optical remote sensing\; and volcanic deform
 ation inversion from Global Navigation Satellite Systems (GNSS) displaceme
 nt data. Across different domains\, PILA yields more accurate and interpre
 table physical variables. For forest spectral inversion\, it improves the 
 separation of tree species and\, compared to ground measurements\, reduces
  prediction errors by 40-71\\% relative to the state-of-the-art. For volca
 nic deformation\, PILA's recovery of variables captures a major inflation 
 event at the Akutan volcano in 2008\, and estimates source depth\, volume 
 change\, and displacement patterns that are consistent with prior studies 
 that however required substantial additional pre-processing. Finally\, we 
 analyse the effects of model rank\, observability\, and physical priors\, 
 and suggest that PILA may offer an effective general pathway for inverting
  incomplete physical models even beyond the domain of Earth Observation. T
 he code is available at https://github.com/yihshe/PILA.git. \n\n*Bio*\n\nY
 ihang She is a third-year PhD student in Computer Science at the Universit
 y of Cambridge\, supervised by Prof. Srinivasan Keshav and Prof. Andrew Bl
 ake. His research focuses on computer vision for forest monitoring and int
 erpretable Earth observation. 
LOCATION:Room GS15 at the William Gates Building and on Zoom: https://cl-c
 am-ac-uk.zoom.us/j/4361570789?pwd=Nkl2T3ZLaTZwRm05bzRTOUUxY3Q4QT09&from=ad
 don 
END:VEVENT
END:VCALENDAR