Daniel Condon is a third-year student finishing his Bachelor of Science in Analytics, with a major in financial mathematics. He has completed subjects in a range of fields including
mathematics, statistics and data science, where he has undertaken a number of data analytics projects. He has a keen interest in using mathematical and statistical tools to solve problems in data science such as improving machine learning techniques.
Applications of Artificial Neural Networks on Structured Datasets
Artificial Neural Networks (ANNs) have become extremely effective in classification tasks involving large unstructured data sets such as in image recognition and natural language processing, however they are still outperformed by other machine learning techniques involving structured data. This project aims to explore Artificial Neural Networks in the context of structured datasets. The research component can be broken down into two sub-components which focus on the following questions:
- Can ANNs outperform traditional machine learning methods on structured datasets?
- How small can the structured dataset be before ANNs become ineffective?
Modern techniques such as entity embedding will be applied to the design of ANNs with the intent of creating an ANN which outperforms traditional machine learning techniques on structured data. Furthermore, this project will explore various sized datasets in an attempt to understand how small a structured dataset must be before ANNs become ineffective.