BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:TESSERA: Temporal Embeddings of Surface Spectra for Earth Represen
 tation and Analysis - Frank Feng\, University of Cambridge
DTSTART:20251114T130000Z
DTEND:20251114T140000Z
UID:TALK237445@talks.cam.ac.uk
CONTACT:114742
DESCRIPTION:*Abstract*\n\nOptical satellite Earth-observation (EO) time se
 ries are often obscured by clouds\, resulting in sparse and temporally irr
 egular observations. Compositing addresses these issues\, but is insensiti
 ve to changes in vegetation phenology\, which is critical for downstream t
 asks. Instead\, we present TESSERA\, a pixel-wise foundation model for mul
 ti-modal (Sentinel-1/2) EO time series that learns robust\, label-efficien
 t embeddings. During model training\, TESSERA uses Barlow Twins to enforce
  invariance to the choice of cloud-free observations randomly sampled from
  the time series\, so that the generated embeddings interpolate missing ob
 servations. We employ two key regularizers: global shuffling to decorrelat
 e spatial neighborhoods\, and mix-based regulation to improve invariance u
 nder extreme sparsity. We find that for diverse classification\, segmentat
 ion\, and regression tasks\, TESSERA embeddings deliver state-of-the-art a
 ccuracy with high label efficiency\, often requiring only a tiny task head
  and minimal computation. To democratize access\, adhere to FAIR principle
 s\, and to simplify use\, we release global\, annual\, 10m\, pixel-wise in
 t8 embeddings together with open weights/code and lightweight adaptation h
 eads\, thus providing practical tooling for large-scale retrieval and infe
 rence at planetary scale.\n\n\n*Bio*\n\nFrank Feng is a second-year Ph.D. 
 student in the Department of Computer Science and Technology at the Univer
 sity of Cambridge. His research interests lie at the intersection of machi
 ne learning and earth sciences\, with a particular focus on the applicatio
 n of self-supervised learning in remote sensing.
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