Domain adaptation has become a prominent problem setting in machine learning and related fields. This review asks the question: how can a classifier learn from a source domain and generalize to a target domain? We present a categorization of approaches, divided into, what we refer to as, sample-based, feature-based and inference-based methods. Sample-based methods focus on weighting individual observations during training based on their importance to the target domain. Feature-based methods revolve around on mapping, projecting and representing features such that a source classifier performs well on the target domain and inference-based methods incorporate adaptation into the parameter estimation procedure, for instance through constraints on the optimization procedure. Additionally, we review a number of conditions that allow for formulating bounds on the cross-domain generalization error. Our categorization highlights recurring ideas and raises questions important to further research.

Accurate prediction of ship trajectories is crucial for ensuring safe and efficient navigation. However, predicting ship trajectories in complex and dynamic environments presents significant challenges. Ships exhibit multi-mode motions, manifesting as diverse motion patterns even under similar circumstances, influenced by factors such as navigational intentions and operational tasks. Moreover, trajectory prediction is further complicated by time-varying ship dynamics, encompassing sailing conditions, ship maneuvering, and environmental factors. In this paper, we propose a Bayesian multiple model with an online model selection strategy to dynamically represent the latent motion mode from early observations. Each sub-model integrates a variational Kalman filter and Gated Recurrent Unit (GRU) neural network, enabling the estimation of time-varying transition coefficients and the process noise specific to different motion modalities. This hybrid methodology leverages the strengths of probabilistic recursive estimation of the Kalman filter while benefiting from the capacity of a GRU network to learn complex temporal dependencies from historical data. The proposed method was evaluated on ship trajectories across different observation lengths and prediction horizons and outperformed the baseline in terms of both accuracy and plausibility.

In machine learning, if the training data is an unbiased sample of an underlying distribution, then the learned classification function will make accurate predictions for new samples. However, if the training data is not an unbiased sample, then there will be differences between how the training data is distributed and how the test data is distributed. Standard classifiers cannot cope with changes in data distributions between training and test phases, and will not perform well. Domain adaptation and transfer learning are sub-fields within machine learning that are concerned with accounting for these types of changes. Here, we present an introduction to these fields, guided by the question: when and how can a classifier generalize from a source to a target domain? We will start with a brief introduction into risk minimization, and how transfer learning and domain adaptation expand upon this framework. Following that, we discuss three special cases of data set shift, namely prior, covariate and concept shift. For more complex domain shifts, there are a wide variety of approaches. These are categorized into: importance-weighting, subspace mapping, domain-invariant spaces, feature augmentation, minimax estimators and robust algorithms. A number of points will arise, which we will discuss in the last section. We conclude with the remark that many open questions will have to be addressed before transfer learners and domain-adaptive classifiers become practical.

Ship collision accidents frequently cause casualties and significant property losses. These collisions mainly occur by incorrectly interpreting the intents of other ships’ navigators. However, inferring the avoidance intents of other ships is challenging, because the uncertain motions and the long-lasting dynamic interactions during encounters usually obscure the true intents. To address this, we propose a probabilistic graph model to infer the hidden avoidance intent of ships. Specifically, the dynamic encounter evolution is expressed in a model represented by a factor graph, where the intent beliefs are accumulated and propagated over time through message passing in the graph. In the inference procedure, we develop a hybrid Bayesian inference approach, integrating a data-driven component derived from empirical priors fitted to historical data, and a model-driven component capturing the ship’s control process. We derive 9 types of intents from naturalistic ship encounters and evaluate the model on a validation split. The quantitative metrics demonstrate that, on average, the proposed procedure can accurately infer the intents 14.04 seconds in advance and outperforms the baseline in macro-averaged recall rate (0.2919) and F1-score (0.2843). Note to Practitioners —This work aims to enhance safety in ship navigation by introducing an intelligent navigation support system to prevent collisions. A significant cause of ship collision accidents is the misinterpretation of other navigators’ intents. Estimating the intents of marine traffic participants can significantly reduce collision rates. However, real-time intent inference in ship encounters poses challenges due to complex dynamics and uncertain motions. We propose a novel probabilistic graphical model that allows us to infer avoidance intents, evaluated on a naturalistic Automatic Identification System (AIS) dataset, demonstrating effectiveness in real-world scenarios. The practical implications of our work are substantial for the maritime industry. Ship captains and intelligent automated vessels can use our system to enhance situational awareness, enabling informed collision avoidance decisions. Our work aims to improve the safety and efficiency of ship navigation, reducing collision risks and ensuring the well-being of crew members and valuable cargo.

Active Inference (AIF) is a framework that can be used both to describe information processing in naturally intelligent systems, such as the human brain, and to design synthetic intelligent systems (agents). In this paper we show that Expected Free Energy (EFE) minimisation, a core feature of the framework, does not lead to purposeful explorative behaviour in linear Gaussian dynamical systems. We provide a simple proof that, due to the specific construction used for the EFE, the terms responsible for the exploratory (epistemic) drive become constant in the case of linear Gaussian systems. This renders AIF equivalent to KL control. From a theoretical point of view this is an interesting result since it is generally assumed that EFE minimisation will always introduce an exploratory drive in AIF agents. While the full EFE objective does not lead to exploration in linear Gaussian dynamical systems, the principles of its construction can still be used to design objectives that include an epistemic drive. We provide an in-depth analysis of the mechanics behind the epistemic drive of AIF agents and show how to design objectives for linear Gaussian dynamical systems that do include an epistemic drive. Concretely, we show that focusing solely on epistemics and dispensing with goal-directed terms leads to a form of maximum entropy exploration that is heavily dependent on the type of control signals driving the system. Additive controls do not permit such exploration. From a practical point of view this is an important result since linear Gaussian dynamical systems with additive controls are an extensively used model class, encompassing for instance Linear Quadratic Gaussian controllers. On the other hand, linear Gaussian dynamical systems driven by multiplicative controls such as switching transition matrices do permit an exploratory drive.