top of page

CATCH – GPT of the sea

A new generation of fish prediction models: AI that understands cause from behavior, not just correlation.

New methodology in the fisheries sector

"No peer-reviewed method has previously shown this kind of implementation in the fisheries sector."

"This is the first published ConvLSTM model for fishing area and catch probability."

CATCH is the first peer-reviewed ConvLSTM AI model that bases fishing area predictions on spatial and temporal patterns from large datasets. It is a new approach to fishing advice, and early results show significantly improved stability, accuracy, and traceability in predicting where fish appear – and why.

SG TMB.jpg

"Just as GPT learns human language from billions of words, CATCH learns the 'language of fish.'"

CATCH reads behavioral patterns: how fish disperse, respond to changes in temperature, depth, and other environmental variables – and then predicts the most likely distribution with scientific accuracy.

Name, Title

How CATCH works

The model is based on ConvLSTM deep learning, which is specifically designed to learn patterns that change in both space and time. By feeding the model with years of environmental and fishing data, it is able to form the "language of fish" - rules and contexts that the human mind cannot directly perceive.

Instead of simple point forecasts, CATCH produces probability distributions that allow users to see uncertainty, fluctuations, and scenarios. This makes it a useful tool for both governments and fishing operators.

Why this matters

In fisheries, details matter. Whether a stock moves 20 km east or 20 km north can have a significant economic impact of tens of billions. CATCH is designed to capture these details with a data-driven precision never before seen in peer-reviewed fisheries methodologies.

With clear probability distributions, traceable forecasts, and continuous verification, CATCH offers a new kind of fish forecast: forecasts that both experts and non-experts can trust.

+354 8935055

  • facebook
  • twitter

©2020 by Blue Economy. 

bottom of page