Article archive
The blog archive now uses the same article system as Articles.
Hierarchical and temporal causation: what changes when causes live at different scales?
Why causal models need to represent both nested levels of abstraction and mechanisms that unfold over time.
Species signatures and a sampling
Representation learning and sampling policy should be designed together so species models generalise across vessels and review resources are spent where they change evidence quality.
From detector accuracy to reviewable fisheries evidence
Hierarchical AI, cross-vessel evaluation across the Pacific, and review-time compression in operational fisheries electronic monitoring.
Fish detection is not enough: evidence for ETP interactions
ETP monitoring needs event reconstruction, rare-event evaluation, independent audit coverage and evidence provenance, not only species accuracy.
What commercial fisheries need from electronic monitoring
A commercial EM product needs evidence capture, review, traceability, standards mapping and measurable workflow value—not a dashboard around a detector.
Electronic monitoring data requirements: what to collect, why it matters, and what AI changes
A system-level view of EM data: cameras, sensors, event schemas, retention, review design and the metrics needed to show real monitoring impact.
Observer data and eLogbooks: the evidence layer before AI
Why observer records and eLogbooks remain the backbone of fisheries evidence, and how to design them for validation, traceability and AI-assisted review.
Dark shipping: absence of AIS is evidence, not a verdict
How to assess missing vessel signals using reception coverage, satellite observations, identity checks and explicit uncertainty.
A unifying perspective on causal world models
A unifying structural framework mapping visual world models to Pearl's Causal Hierarchy of Seeing, Interpreting, Doing, and Imagining.
Why a 99% maritime anomaly score may tell you very little
A practical evaluation checklist for AIS anomaly models: label provenance, synthetic events, shared test sets, leakage, class definitions and independent evidence.
Prediction is not intervention: a causal view of port-call optimisation
Why port-call optimisation should move from ETA prediction to causal intervention modelling over vessel, berth, weather, cargo and stakeholder decisions.
Steering optimisation trajectories in diffusion representation learning
Analysing and steering optimisation dynamics reconciles the trade-off between latent disentanglement and pixel reconstruction in diffusion autoencoders.
CO-ALIGN: bias mitigation in diffusion models using Concept Graphs
CO-ALIGN aligns concept graphs between text encoders and visual denoisers to eliminate demographic bias in text-to-image diffusion models.
Transshipment detection should model an interaction, not two nearby tracks
From proximity thresholds to probabilistic interaction modelling for fishing-vessel and carrier encounters.
Emergent tactics in Formula One race strategy via reinforcement learning
RSRL connects real-time telemetry and proprietary simulators to learn emergent pitstop tactics in multi-agent Formula One races.
What Does Port Congestion Actually Measure?
Why congestion should be modelled through service backlog, effective capacity and operational constraints rather than as a count of ships inside an anchorage.
Loitering, waiting and anchorage are different states
Why maritime loitering detection should distinguish anchorage, operational waiting, drifting and anomalous local motion before training a classifier.
Anchorage detection is a behaviour-inference problem
Why anchorage detection should be framed as latent-state inference over imperfect AIS rather than as a speed threshold inside a polygon.
Maritime sensing is an observability problem: AIS, EO, SAR and RF
What vessel broadcasts, satellite images and radio detections can establish, how to combine them, and where uncertainty remains.
TRuST: selective fine-tuning for robust concept unlearning in diffusion models
TRuST dynamically identifies concept neurons and applies Hessian-regularised selective fine-tuning to erase harmful concepts from diffusion models.
Object-centric case-based reasoning via argumentation
SAA-CBR combines Slot Attention with Abstract Argumentation for Case-Based Reasoning to classify images through precedent-based dialectical reasoning.
Transparent visual reasoning via object-centric agent negotiation
OCEAN formulates visual reasoning as a multi-agent negotiation game where agents select and debate object explananda to reach transparent classifications.
Causal representation learning with observational grouping for chest X-rays
Observational grouping enforces causal invariance across demographic and scanner environments to eliminate diagnostic shortcuts in chest X-rays.
Neuro-argumentative learning: uniting object slots with symbolic logic
Neural Argumentative Learning (NAL) marries Slot Attention with Assumption-Based Argumentation for transparent, rule-grounded image classification.
Resolving spatial ambiguities in object-centric representations
View-Invariant Slot Attention (VISA) resolves partial occlusions and multi-view ambiguities with theoretical identifiability guarantees.
Diffusion counterfactuals with semantic abduction
Integrating semantic representations into diffusion models through Pearl's causal ladder enables identity-preserving counterfactual image generation.
Towards human-like visual reasoning: an object-centric foundation
A comprehensive synthesis of object-centric representation learning, theoretical identifiability, and causal reasoning.
Causal representation learning via mixture-based priors
Partitioning the latent space into causal factors and exogenous noise enables identifiable causal representation learning without interventional data.
Causal composition in object-centric representations
Connecting object-centric learning with causal representation learning enables causally consistent scene generation and composition.
Formula One pitstop strategy via explainable reinforcement learning
Race Strategy Reinforcement Learning (RSRL) optimizes tyre compound choices and pitstop laps in Formula One simulations.
Free Argumentative Exchanges: multi-agent debates for visual explanations
Free Argumentative Exchanges (FAXs) enable agents to debate and internalise visual evidence using Bipolar Argumentation Frameworks.
Probabilistic Slot Attention: learning identifiable object representations
Probabilistic Slot Attention (PSA) introduces a generative formulation over object slots, proving identifiability without supervised masks.
Continuous Bayesian model selection for multivariate causal discovery
A continuous relaxation of Bayesian model selection identifies multivariate directed acyclic graphs without restrictive additive noise assumptions.
Visual debates: explaining image classifiers through dialogue
Visual debates replace static saliency heatmaps with interactive argumentative dialogue between two agents.
Grounded Object-Centric Learning: giving slots a reusable vocabulary
Conditional Slot Attention learns a Grounded Slot Dictionary for reusable object representations.
Learning hidden variables in a causal order
The paper studies representations whose hidden variables can influence one another.
Explaining an image classifier through a hierarchy of symbols
The method explains a trained image classifier using a learned hierarchy of discrete symbols.
LiTS: why a good liver outline can still miss a tumour
LiTS evaluates liver and tumour segmentation on a shared collection of abdominal CT scans.