<span class="translation_missing" title="translation missing: en.options.talk">talk</span>: Keeping waterways clean with Machine Learning

How might we use machine learning to reduce time spent on manual review while accurately identify-ing lapses?

We addressed this challenge with the Singapore Public Utilities Board (PUB), whose task is it to make sure that any constructions site’s Earth Control Measure (ECM) submission adheres to the necessary re-quirements, to ensure the site poses no damage to the environment.

We are going to dive into how we tackled this problem, together with the learnings we had during the project.

The Singapore Public Utilities Board (PUB) is responsible for the management of the national water supply, water catchment, and used water. As such, one of their tasks is to make sure that any construction site’s Earth Control Measure (ECM) submission adheres to the necessary requirements, to ensure the site poses no damage to the environment. PUB receives more than 1’500 drawing submissions per year seeking approval for construction works that require ECM to be implemented. This process requires time as well as an experienced eye to manually review the drawings and ensure that the ECMs are properly designed. PUB asked Zühlke to assess the feasibility of automating the process of checking ECM plans through Machine Learning. Over a span of four months, the team built a corresponding end-to-end prototype in the cloud which is trained to decipher past submissions and identify commonly spotted issues in submissions to optimise the overall process. In this talk we present the approach we chose to tackle this problem, together with the main challenges and learnings we had during the project.