Classifying Shark Behavior in Time and Frequency Assignment

Assignment Task

Project 5 Shark Behavior Classification

A time-series data for sharks is available at Canvas.

We want to use known acceleration lab data to build a model of behavior signatures by “classification”, including:

1. Resting

2. Swimming

3. Feeding

4. Non-directed motion (NDM)

Compare the classification capability of the following neural networks on 7 shark data sets of each category:

1. NN

2. LSTM or RNN

3. K-NN (done, see Yeh’s papers)

You need to consider the performance of each category data ranges of time from t1 into the future t2, say

The main features of the data set are:

• High resolution (sampling frequency = 25 Hz) for x, y, z axis (3-D dynamic acceleration) and rotation on x, y, x axis (3-D Static acceleration)
• Biased data sets. “Feeding” data is much smaller than other categories (i.e., Resting, Swimming, NDM)

This Management has been solved by our PHD Experts at My Uni Paper.