Machine Learning and Deep Learning for Ecologists (Kannan AS): day4

“””With an increased focus on climate change and biodiversity, there is a lot of ecological data available from varied sources like citizen science, bird/mammal surveys, and camera trap images. Traditional statistical methods are limited in handling such a variety and quantum of data. Statistical modelling is about finding patterns and relationships between variables and the significance of those patterns. Today’s ecological data has a lot of variables and finding specific patterns across many these variables is very difficult with traditional statistical modelling. Machine learning models not only help with clustering the sample data over many variables but can predict an outcome given a new example data. Secondly the availability of multimedia data like acoustics, images and video opens a new set of challenges that can be addressed quite easily with Deep learning algorithms!
Despite these advantages of Machine / Deep learning models, there has been a slow adoption of these techniques in ecology. This could be addressed by increasing the awareness of these techniques among ecologists and by building collaboration between ecologists and machine learning / deep learning community.
This workshop aims to provide a simple introduction to Machine learning and deep learning models and a few potential applications in Ecology that can be considered. We will also highlight a few interesting applications from across the world to provide a flavor of the power of ML/DL. At NCF we have used Deep learning for classification of animals from camera trap images. We will provide the methodology used and its effectiveness along with a demo of the model in action!”””