\section{Conclusion} \label{sec:conclusion} We presented **MarVelocity**, a hybrid metric that blends classical hydrodynamic resistance modelling with a universal machine‑
\begin{figure}[H] \centering \includegraphics[width=0.75\linewidth]{ablation.png} \caption{Ablation results: MAE increase when a feature group is omitted.} \label{fig:ablation} \end{figure} marvelocity pdf
\section{Related Work} \label{sec:related} \subsection{Physical Models} The Holtrop–Mennen (HM) and KVLCC2 families remain industry standards for estimating ship resistance \cite{Holtrop1972, KVLCC1992}. Their primary limitation is the assumption of steady, uniform sea conditions and neglect of wind‑induced drag. \end{itemize} Model training is performed on a single
The final **MarVelocity** prediction is: \begin{equation} V_{\text{MarV}} = V_{\text{HM}} + \hat{\Delta V}(\mathbf{x}). \end{equation} Massachusetts Institute of Technology
\subsection{Training Procedure} \begin{itemize} \item \textbf{Train/validation split}: 70 \% ships for training, 15 \% for validation, 15 \% for test (no ship appears in more than one split). \item \textbf{Hyper‑parameter optimisation}: Bayesian optimisation (Optuna \cite{Akiba2019}) over tree depth, learning rate, and number of estimators. \item \textbf{Loss function}: Mean Absolute Error (MAE) on $\Delta V$. \end{itemize} Model training is performed on a single NVIDIA RTX 4090 GPU (≈ 5 min).
\title{MarVelocity:\\A Data‑Driven Metric for Predicting Maritime Vessel Speed} \author{ \textbf{Alexandra T. Liu}$^{1}$, \textbf{Rahul K. Menon}$^{2}$, \textbf{Elena G. Petrova}$^{3}$\\[2mm] $^{1}$Department of Naval Architecture, Massachusetts Institute of Technology, Cambridge, MA, USA\\ $^{2}$Marine Systems Research Group, Indian Institute of Technology, Bombay, India\\ $^{3}$Institute of Ocean Engineering, Technical University of Munich, Munich, Germany\\[2mm] \texttt{atl@mit.edu, rkm@iitb.ac.in, elena.petrova@tum.de} } \date{\today}
