Researchers create fast explainable AI for landslide detection by cutting data layers
Graduate student Arsalaan Ahmad and his mentors published a new method for detecting landslides using satellite AI in the journal Frontiers in Remote Sensing.
New approach reduces model size
The research team addressed the black box problem found in current artificial intelligence models. They achieved this by cutting data layers to make the system faster and easier to explain. This technique removes up to 30 layers from standard models without losing accuracy.
Author background and inspiration
Lead author Arsalaan Ahmad grew up in Oman where he studied mountain geography. He developed an interest in preventing natural disasters after seeing construction measures in his home country. The 23-year-old student recently graduated from Cardiff University with a degree in computer science.
Collaboration and publication details
Dr Oktay Karakuş and Professor Paul Rosin co-authored the paper alongside Ahmad. The group published their findings in a top-tier academic journal focused on remote sensing. Their work aims to improve how authorities map natural disasters using satellite data.
Reported by one outlet
Only one outlet has published this. Nothing here has been checked against a second report, so read it as that outlet's account and follow the link below for the original.
Reported by
1 independent outlet. Headline as published. Links open the original report.