Data to the Rescue! Predicting and Preventing Accidents at Sea | Dr. Yonit Hoffman

Discover how data science and AI can predict and prevent accidents at sea, improving safety and reducing environmental, financial, and human impacts, while also increasing transparency and trust in machine learning models.

Key takeaways

Data science and AI can be used to predict and prevent accidents at sea.
Machine learning models can gain trust by being transparent about their decision-making process.
SHAP values can be used to explain the reasoning behind a model’s predictions.
By analyzing time series data, predictions can be made about a ship’s speed, direction, and location.
Transferring learning from one model to another can improve performance.
Explainability is key to gaining trust in machine learning models, especially in traditional industries.
Data scientists and users need to work together to understand the risk of a group of ships.
Accidents at sea can have significant environmental, financial, and human impacts.
Machine learning can be used to detect risky owners, ships, and behaviors.
Predictive models can be used to identify potential accidents, such as grounding or collision.
Transferring learning from one model to another can improve performance.
Explainability is key to gaining trust in machine learning models.
Data scientists and users need to work together to understand the risk of a group of ships.
Accidents at sea can have significant environmental, financial, and human impacts.
Machine learning can be used to detect risky owners, ships, and behaviors.
Predictive models can be used to identify potential accidents, such as grounding or collision.
The Windward system uses XGBoost and TCN models to predict accidents at sea.
Insurance companies are interested in the risk of the group of ships, not just the probability of an accident.
Machine learning models need to be transparent about their decision-making process to gain trust.
Data scientists can out front in creating machine learning-based products.
Explainability is key to gaining trust in machine learning models, especially in traditional industries.
Data scientists and users need to work together to understand the risk of a group of ships.
Accidents at sea can have significant environmental, financial, and human impacts.
Machine learning can be used to detect risky owners, ships, and behaviors.
Predictive models can be used to identify potential accidents, such as grounding or collision.
The Windward system provides a platform for data scientists to work on maritime data.
Data science and AI can be used to predict and prevent accidents at sea.
Machine learning models need to be transparent about their decision-making process to gain trust.
Data scientists and users need to work together to understand the risk of a group of ships.
Predictive models can be used to identify potential accidents, such as grounding or collision.

Data to the Rescue! Predicting and Preventing Accidents at Sea | Dr. Yonit Hoffman

More talks