Speaker
Description
Acoustic telemetry data plays a vital role in understanding the be-
haviour and movement of aquatic animals. However, these datasets,
which can often consist of millions of individual data points, often
contain anomalous detections that can pose challenges in data analysis
and interpretation. Anomalies in acoustic telemetry data can occur due
to various biological and environmental factors, and technological limi-
tations. Anomalous movements are generally identified manually, which
can be extremely time-consuming in large datasets. As such, this study
focuses on automating the process of anomaly detection in telemetry
datasets using machine learning (ML) and artificial intelligence (AI)
models. Fifty dusky kob (Argyrosomus japonicus) were surgically fit-
ted with unique coded acoustic transmitters in the Breede Estuary,
South Africa, and their movements were monitored using an array of
16 acoustic receivers deployed throughout the estuary between 2016
and 2021, resulting in more than 3 million individual data points. The
research approach combined the use of Neural Network (NN) models
and autoencoders to construct an efficient anomaly detection system. The model is proficient at learning the normal movement patterns within
the data, effectively distinguishing between normal and anomalous be-
haviour, and exceeding 90% across all four evaluation metrics including
accuracy, precision, recall, and F1. However, it may encounter chal-
lenges in accurately detecting anomalies where they deviate slowly from
the expected movement patterns. Despite this limitation, the model
demonstrates promising capabilities by pinpointing the precise loca-
tions of anomalous entries within the dataset. Further investigation,
including refinement and optimization of the model’s parameters and
training process, especially with memory-based NN-AE, may enhance
its ability to detect anomalies with greater accuracy and reliability.
