Presenting at the European Council on Computing in Construction 2024
On 15/07/2024, I had the opportunity to attend the European Council on Computing in Construction 2024 conference at Crete, Greece, where I presented our team’s latest research: “Estimate Road Roughness Using Smartphone Response Data - A Semi-Supervised Learning Approach”. This work was co-authored with Ye Sang, Dr. Yihai Fang, Dr. Viet Vo, and Richard Wix.
Tackling the Challenge of Road Roughness Estimation
Developing a reliable, data-driven International Roughness Index (IRI) estimation model is often costly due to the extensive labelled data required, typically collected through conventional survey vehicles. However, the widespread availability of unlabelled response data from road users opened up new possibilities for us.
We explored the potential of semi-supervised learning (SSL) algorithms, which enabled us to utilise both labelled and unlabelled data effectively. Our study showed that by applying SSL, we improved the IRI estimation model and achieved a lower Root Mean Squared Error (RMSE) compared to the baseline model, which uses the same amount of labelled data.
text
Photos from the Conference
