My team and I graduated from a data science master's program from UC Berkeley in May 2021, where we collaborated together on our capstone project, which was about the proposed topic. By trade, we have experience in software engineering and data science, with one of our team members being our maritime domain expert, given his extensive experience in the US Coast Guard.
Quick Summary
Our team went into this project with the goal of leveraging publicly available, historical data on maritime navigation patterns and standard deep learning techniques inspired by the realm of natural language processing to more effectively tackle the problem of predicting vessels trajectories and improving maritime domain awareness.
For this lesson, our goal is to guide our audience through our proposed technique in an intuitive and visualized way to really help people understand the problem space of vessel trajectory prediction and understand how we even got to our proposed solution. We hope to use this lesson as an opportunity to celebrate the kind of innovation that can come from the collaboration of unique fields like those of data science and maritime navigation.
Target medium
We are open to most any medium that helps to best explain our technique, but at this time, we have only seriously considered video-based explanations and article-based explanations. Video explanations along the lines of those demonstrated by the YouTube channel AI Coffee Break is an example of something that might work for us, but we're happy to do some exploration here.
More details
We have an high-level website that walks through the scenario and the applications of the technique here: www.rule5.ai (Note that there is a YouTube video linked at the bottom of the home page with details and a walk through of our MVP for the capstone project).
Contact details
If you're interested or have questions, please reach out to us via dianai@berkeley.edu.
Additional context
We are not aiming to hit the August deadline proposed by Grant for this event, but wanted to see if there are content producers who might be interested in helping us develop an explanation of our technique. Would appreciate content producers to be decently versed in basics of neural networks, but this is not required.
About the author(s)
My team and I graduated from a data science master's program from UC Berkeley in May 2021, where we collaborated together on our capstone project, which was about the proposed topic. By trade, we have experience in software engineering and data science, with one of our team members being our maritime domain expert, given his extensive experience in the US Coast Guard.
Quick Summary
Our team went into this project with the goal of leveraging publicly available, historical data on maritime navigation patterns and standard deep learning techniques inspired by the realm of natural language processing to more effectively tackle the problem of predicting vessels trajectories and improving maritime domain awareness.
For this lesson, our goal is to guide our audience through our proposed technique in an intuitive and visualized way to really help people understand the problem space of vessel trajectory prediction and understand how we even got to our proposed solution. We hope to use this lesson as an opportunity to celebrate the kind of innovation that can come from the collaboration of unique fields like those of data science and maritime navigation.
Target medium
We are open to most any medium that helps to best explain our technique, but at this time, we have only seriously considered video-based explanations and article-based explanations. Video explanations along the lines of those demonstrated by the YouTube channel AI Coffee Break is an example of something that might work for us, but we're happy to do some exploration here.
More details
We have an high-level website that walks through the scenario and the applications of the technique here: www.rule5.ai (Note that there is a YouTube video linked at the bottom of the home page with details and a walk through of our MVP for the capstone project).
Contact details
If you're interested or have questions, please reach out to us via dianai@berkeley.edu.
Additional context
We are not aiming to hit the August deadline proposed by Grant for this event, but wanted to see if there are content producers who might be interested in helping us develop an explanation of our technique. Would appreciate content producers to be decently versed in basics of neural networks, but this is not required.
Licensing: CC-BY