How We Used IBM Cloud for the FishAI Competition


Commercial fishing vessel and sea birds flying around it sunset time.

We participated in FishAI: Sustainable Commercial Fishing Competition. It was organized by NORA (Norwegian Artificial Intelligence Research Consortium) and many other collaborators, and it aimed to find solutions to make commercial fishing sustainable utilizing artificial intelligence. Intito’s team ended up being in the top 5 of nearly 40 teams who participated in the FishAI competition.

Optimal Fishing Routes Are Forecasted Based on Historical Data

With a fishing zone of 2.1 million square meters, Norway is considered Europe’s largest fishing and aquaculture nation. The organizers provided a large dataset of various data related to fishing from over 20 years. Our solution forecasts where a commercial fishing vessel most likely will catch certain type of fish on a specific day based on the provided historical data.

Most of the forecasted optimal locations for fishing that our solution suggests are near the costal line of Norway. For some species the forecasted optimal fishing route was quite stationary and for others it suggested a lot more moving around.

Our solution included the visualization of the fishing routes, how much fish could be expected, the length of the fishing routes in kilometers and calculations of fuel consumption. For the forecasting we tested regression and classification models. To create optimal fishing routes, we used a clustering algorithm and a mathematical optimization model.

Developing a Comprehensive Data Science Solution on IBM Cloud

For us the competition was a fun and motivating starting point for building a comprehensive data science demo solution on IBM Cloud. During the competition we ended up using Auto AI, Jupyter Notebooks, DB2 database, Cloud Object Storage and Decision Optimization. Everything is stored in IBM Cloud.

“We will continue working on this demo solution. These same forecasting and optimization methods could be used in a completely different use case, for example predicting mail order locations and optimizing delivery. We will develop the visual reports of the solutions further with Cognos Dashboards reporting capabilities. Overall, participating in the competition was a good way to start developing a comprehensive data science solution, but we still have a lot to do to develop it further and make it better!” Visa Linkiö, Lead Data Scientist, Intito concludes.

A deep dive into this project can be watched as an on-demand webinar.

If you want to read more about the competition, check out their website!