Related Experiment Videos
Classification of drug-induced behaviors using a multi-layer feed-forward neural network
1Department of Psychiatry and Behavioral Sciences, University of Oklahoma Health Sciences Center, Oklahoma City 73190-3000.
Computer Methods and Programs in Biomedicine
|July 1, 1993
Summary
This study introduces a novel signal analysis technique for automatically classifying multiple laboratory animal behaviors. The method enhances the quantitative assessment of motor activity in neuroscience research.
Area of Science:
- Neuroscience
- Behavioral Science
- Signal Processing
Background:
- Accurate measurement of laboratory animal motor behavior is crucial for central nervous system research.
- Existing automated systems have limitations in simultaneously monitoring diverse behaviors.
Purpose of the Study:
- To describe a new signal analysis technique for automatic classification of multiple animal behaviors.
- To overcome limitations of current automated behavioral monitoring systems.
Main Methods:
- Utilized a modified electronic activity monitor and signal analysis.
- Employed fixed-length segmentation, Fourier transform, and power spectrum analysis.
- Implemented an error back-propagation neural network for behavior classification.
Main Results:
- The technique successfully classifies multiple behavior categories automatically.
- Achieved a high degree of accuracy in automatic behavior classification.
Conclusions:
- The described signal analysis technique offers an accurate method for automated behavioral classification.
- This approach is valuable for the quantitative assessment of motor behavior in laboratory animals.