NSF HDR ML Challenge

Scientific discovery often involves finding an inconsistent pattern within our data. Data that behaves differently from what is expected can indicate that the underlying science is different. Different behavior can result from a number of effects, but ultimately this could imply that we have observed something new 🌟!

Depending on the scientific domain, a new, unpredictable object/event could have a profound impact. This could be a new type of material, the discovery of a new astrophysical object 🌌, the observation of unusual climate behavior 🌦️, or the discovery of a new species 🦋. The observation of something different, incongruous with the data, is what we call anomaly detection 🔍. Looking for anomalies is often quite different than other tasks since we do not know what exactly to look for, we just need to look for something different.

The main focus of this challenge is the application of machine learning to scientific anomaly detection 🤖.

The ML Challenge is extended to run through January 31st, 2025!

Participate in the Anomaly Detection challenge

Challenge Organizers

Imageomics

  • Elizabeth G. Campolongo
  • Wei-Lun Chao
  • Hilmar Lapp

A3D3

  • Yuan-Tang Chou
  • Ekaterina Govorkova
  • Philip Harris
  • Shih-Chieh Hsu
  • Mark S. Neubauer

iHarp

  • Aneesh Subramanian
  • Josephine Namayanja
  • Vandana Janeja
  • Shashi Shekhar

Student Organizers

Imageomics

  • Jiaman Wu
  • David E. Carlyn
  • Christopher Lawrence
  • Ziheng Zhang

A3D3

  • Advaith Anand
  • Eric Moreno
  • Ryan Raikman

iHarp

  • Subhankar Ghosh
Butterfly
Discovering hybrid butterfly species through pattern recognition in image datasets
GW
Finding unmodeled gravitational wave events, such as potential supernovae, in detector data
Water Levels
Identifying unusual fluctuations in water levels that do not correspond to known environmental factors or historical data patterns