As technology makes jobs obsolete - from the gas station attendant to the VCR repairman to the travel agent - a new article in the journal Ecological Informatics foretells a possible future scenario in which the position of "research assistant" sits next-on-the-chopping-block.
Eric Kasten and fellow scientists from the Remote Environmental Assessment Laboratory at Michigan State University have developed and tested an automated system for remotely surveying birds based on their vocalizations.
The system involves using pole-mounted, remote sensors to collect ongoing acoustic data. These continuous data streams are automatically processed across data pipelines to distill meaningful acoustic sequences - what the authors term 'Ensembles."
A program called MEMO - a perceptual memory system - then analyzes these ensembles to identify the species present. While this automated approach was developed specifically to monitor birds, the method is likely applicable to other groups of animals that are subject to acoustic surveys.
The researchers tested the system across 10 avian species and found a fairly high level of accuracy - over 70% - in classifying the birds. They note that combining this acoustical data with other environmental information (e.g. temperature, humidity, etc) might further enhance the predictive capacity of the system.
Thinking further into the future, one can imagine an automated ecological monitoring system that integrates a wide variety of biological and physical data streams to assess the health of ecosystems in real time. The authors write,
"Monitoring the health of an ecosystem will require the acquisition and correlation of data from many sensors to capture the complex behavior afforded by multiple interacting systems and organisms."
--Reviewed by Rob Goldstein
Kasten, E., McKinley, P., & Gage, S. (2010). Ensemble Extraction for Classification and Detection of Bird Species☆ Ecological Informatics DOI: 10.1016/j.ecoinf.2010.02.003