{
  "id": 18159299,
  "title": "Boffins use machine-learning tricks to create a starlight dimmer switch",
  "url": "https://urgent.news/2026/11/04/boffins-use-machine-learning-tricks-to-create-a-starlight-dimmer-18159299",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-04T18:03:00.000Z",
  "source": {
    "name": "The Register",
    "slug": "the-register",
    "url": "https://www.theregister.com/ai-and-ml/2026/11/04/boffins-use-machine-learning-tricks-to-create-a-starlight-dimmer-switch/5307625"
  },
  "original_language": "en",
  "account": "Scientists have devised a novel method to eliminate starlight from telescope images, enhancing the sensitivity to detect faint exoplanets by as much as fifteen times in trials conducted on James Webb Space Telescope data. Led by Northwestern University's Rodrigo Ferrer-Chavez, a team integrated a physics-based telescope model with machine learning optimization techniques to subtract unwanted starlight. Directly observing exoplanets proves challenging due to the overwhelming glare from their parent stars. While existing methods utilize modeled or reference starlight for subtraction, minor fluctuations in telescope optics can leave residual glare that obscures planets. To address these issues, the researchers developed the Physics-based Learning for Imaging Planets (FLIP) framework. This framework accurately models light passage through the telescope, accounts for optical imperfections, and simulates the glare to be removed. When applied to observations from Webb's Near-Infrared Camera (NIRCam), FLIP yielded the initial image of the known giant planet 51 Eridani b at an infrared wavelength of 4.3 micrometers, a wavelength where traditional processing had failed. The FLIP framework demonstrated sensitivity improvements ranging from two to fifteen times when compared to conventional techniques, as reported in a Nature paper. This marks the first instance, to the authors' knowledge, of such an approach yielding substantial sensitivity enhancements on real, on-sky coronagraphic imaging. The researchers posit that their framework could be adapted for use with other existing and future ground-based and space-based observatories, potentially alleviating the stringent stability requirements necessary for successful direct imaging. Future endeavors will focus on reducing computational demands and devising methods to handle the significantly larger data volumes generated by ground-based observations. The researchers anticipate that further refinement and testing of FLIP will bring direct images of Earth-like planets within reach, while simultaneously mitigating the impact of microscopic telescope instabilities on image clarity.",
  "summary": "FLIP reveals a planet hidden in Webb data, boosting detection sensitivity by up to 15 times",
  "key_points": [
    "Scientists developed FLIP framework to eliminate starlight from telescope images",
    "FLIP improved sensitivity to detect faint exoplanets by 2-15 times",
    "Framework could adapt to ground-based and future space-based observatories"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}