arXiv subjects

Tina Comes

Publications and source records attributed to Tina Comes.

8 recordsLinked to original sources

arXiv2026

Towards Reliable Zero-Shot Crowd Forecasting: Evaluating Time Series Foundation Models for Special Event Pedestrian Forecasting

Managing massive crowds during infrequent special events requires reliable real-time pedestrian-flow forecasting to ensure public safety and operational efficiency. However, supervised forecasting methods face limitations in these contexts due to scarce historical data, heterogeneous data distributions, and short in-event observation windows. To effectively support operational decision-making, forecasts should provide not only accurate point estimates but also informative predictive uncertainty. Probabilistic uncertainty quantification plays a critical role in this aspect, particularly capturing sudden volatility and tail risks. This paper investigates pretrained time series foundation models as a lightweight approach for zero-shot probabilistic forecasting without extensive local retraining. Using decision-oriented metrics tailored to short events, we conduct a comprehensive assessment of two time series foundation models on crowd forecasting, with the SAIL2025 event as a use case. We then distill practical insights for crowd managers, specifying when zero-shot forecasts remain operationally reliable.

cs.LG↗

arXiv2025

A Method for Rapid Area Prioritisation in Flood Disaster Response

In flood disasters, decision-makers have to rapidly prioritise the areas that need assistance based on a high volume of information. While approaches that combine GIS with Bayesian networks are generally effective in integrating multiple spatial variables and can thus reduce cognitive load, existing models in the literature are not equipped to address the time pressure and information-scape that is typical in a flood. To address the lack of a model for area prioritisation in flood disaster response, we present a novel decision support system that adheres to the time and information characteristics of an ongoing flood to infer the areas with the highest risk. This decision support system is based on a novel GIS-informed Bayesian network model that reflects the challenges of decision-making for area prioritisation. By developing the model during the preparedness phase, some of the most time-consuming aspects of the decision-making process are removed from the time-critical response phase. In this way, the proposed method aims to providing rapid and transparent area prioritisation recommendations for disaster response. To illustrate our method, we present a case study of an extreme flood scenario in Cologne, Germany.

stat.AP↗

arXiv2024

Learning to connect in action: Measuring and understanding the emergence of boundary spanners in volatile times

Collective intelligence of diverse groups is key for tackling many of today's grand challenges such as fostering resilience and climate adaptation. Information exchange across such diverse groups is crucial for collective intelligence, especially in volatile environments. To facilitate inter-group information exchange, Informational Boundary Spanners (IBSs) as pivotal information exchange 'hubs' are promising. However, the mechanisms that drive the emergence of IBSs remain poorly understood. To address this gap there is first a need for a method to identify and measure the emergence of IBSs. Second, an Agent-Based Modelling (ABM) framework is not available to systematically study mechanisms for the emergence of IBSs in volatile environments. Third, even though the ability to learn who provides high-quality information is thought to be essential to explain the emergence of IBSs, a rigorous test of this mechanism is missing. The learning mechanism is formalized using an ABM framework, with the model's outputs analyzed using the proposed IBS emergence measurement method. To illustrate both the method and the learning mechanism, we present a case study focused on information sharing in the volatile environment of a disaster. The study shows that learning constitutes a mechanism for the emergence of effective IBSs in (a) low-volatility environments characterised by low uncertainty and (b) in high-volatility environments characterised by rapid change if the number of inter-group connections is sufficient. With the method and model, this paper aims to lay the foundations for exploring mechanisms for the emergence of IBSs that facilitate inter-group information exchange. This article advances collective intelligence by providing the essential elements for measuring and understanding the emergence of IBSs and exploring the effect of learning on their emergence in volatile environments.

cs.MA↗

arXiv2024

Emergency Response Inference Mapping (ERIMap): A Bayesian network-based method for dynamic observation processing

In emergencies, high stake decisions often have to be made under time pressure and strain. In order to support such decisions, information from various sources needs to be collected and processed rapidly. The information available tends to be temporally and spatially variable, uncertain, and sometimes conflicting, leading to potential biases in decisions. Currently, there is a lack of systematic approaches for information processing and situation assessment which meet the particular demands of emergency situations. To address this gap, we present a Bayesian network-based method called ERIMap that is tailored to the complex information-scape during emergencies. The method enables the systematic and rapid processing of heterogeneous and potentially uncertain observations and draws inferences about key variables of an emergency. It thereby reduces complexity and cognitive load for decision makers. The output of the ERIMap method is a dynamically evolving and spatially resolved map of beliefs about key variables of an emergency that is updated each time a new observation becomes available. The method is illustrated in a case study in which an emergency response is triggered by an accident causing a gas leakage on a chemical plant site.

cs.IR↗

arXiv2023

Weaving Equity into Infrastructure Resilience Research and Practice: A Decadal Review and Future Directions

After about a decade of research in this domain, what is missing is a systematic overview of the research agenda across different infrastructures and hazards. It is now imperative to evaluate the current progress and gaps. This paper presents a systematic review of equity literature on disrupted infrastructure during a natural hazard event. Following a systematic review protocol, we collected, screened, and evaluated almost 3,000 studies. Our analysis focuses on the intersection within the dimensions of the eight-dimensional assessment framework that distinguishes focus of the study, methodological approaches, and equity dimensions (distributional-demographic, distributional-spatial, procedural, and capacity equity). To conceptualize the intersection of the different dimensions of equity, we refer to pathways, which identify how equity is constructed, analyzed, and used. Significant findings show that (1) the interest in equity in infrastructure resilience has exponentially increased, (2) the majority of studies are in the US and by extension in the global north, (3) most data collection use descriptive and open-data and none of the international studies use location-intelligence data. The most prominent equity conceptualization is distributional equity, such as the disproportionate impacts to vulnerable populations and spaces. The most common pathways to study equity connect distributional equity to the infrastructure's power, water, and transportation in response to flooding and hurricane storms. Other equity concepts or pathways, such as connections of equity to decision-making and building household capacity, remain understudied. Future research directions include quantifying the social costs of infrastructure disruptions and better integration of equity into resilience decision-making.

physics.soc-ph↗

arXiv2022

Roadmap Towards Responsible AI in Crisis Resilience Management

Novel data sensing and AI technologies are finding practical use in the analysis of crisis resilience, revealing the need to consider how responsible artificial intelligence (AI) practices can mitigate harmful outcomes and protect vulnerable populations. In this paper, we present a responsible AI roadmap that is embedded in the Crisis Information Management Circle. This roadmap includes six propositions to highlight and address important challenges and considerations specifically related to responsible AI for crisis resilience management. We cover a wide spectrum of interwoven challenges and considerations pertaining to the responsible collection, analysis, sharing, and use of information such as equity, fairness, biases, explainability and transparency, accountability, privacy and security, inter-organizational coordination, and public engagement. Through examining issues around AI systems for crisis resilience management, we dissect the inherent complexities of information management and decision-making in crises and highlight the urgency of responsible AI research and practice. The ideas laid out in this paper are the first attempt in establishing a roadmap for researchers, practitioners, developers, emergency managers, humanitarian organizations, and public officials to address important considerations for responsible AI pertaining to crisis resilience management.

cs.SI↗

arXiv2022

On the interplay of data and cognitive bias in crisis information management -- An exploratory study on epidemic response

Humanitarian crises, such as the 2014 West Africa Ebola epidemic, challenge information management and thereby threaten the digital resilience of the responding organizations. Crisis information management (CIM) is characterised by the urgency to respond despite the uncertainty of the situation. Coupled with high stakes, limited resources and a high cognitive load, crises are prone to induce biases in the data and the cognitive processes of analysts and decision-makers. When biases remain undetected and untreated in CIM, they may lead to decisions based on biased information, increasing the risk of an inefficient response. Literature suggests that crisis response needs to address the initial uncertainty and possible biases by adapting to new and better information as it becomes available. However, we know little about whether adaptive approaches mitigate the interplay of data and cognitive biases. We investigated this question in an exploratory, three-stage experiment on epidemic response. Our participants were experienced practitioners in the fields of crisis decision-making and information analysis. We found that analysts fail to successfully debias data, even when biases are detected, and that this failure can be attributed to undervaluing debiasing efforts in favor of rapid results. This failure leads to the development of biased information products that are conveyed to decision-makers, who consequently make decisions based on biased information. Confirmation bias reinforces the reliance on conclusions reached with biased data, leading to a vicious cycle, in which biased assumptions remain uncorrected. We suggest mindful debiasing as a possible counter-strategy against these bias effects in CIM.

cs.CY↗

arXiv2021

A Machine learning approach for rapid disaster response based on multi-modal data. The case of housing & shelter needs

Along with climate change, more frequent extreme events, such as flooding and tropical cyclones, threaten the livelihoods and wellbeing of poor and vulnerable populations. One of the most immediate needs of people affected by a disaster is finding shelter. While the proliferation of data on disasters is already helping to save lives, identifying damages in buildings, assessing shelter needs, and finding appropriate places to establish emergency shelters or settlements require a wide range of data to be combined rapidly. To address this gap and make a headway in comprehensive assessments, this paper proposes a machine learning workflow that aims to fuse and rapidly analyse multimodal data. This workflow is built around open and online data to ensure scalability and broad accessibility. Based on a database of 19 characteristics for more than 200 disasters worldwide, a fusion approach at the decision level was used. This technique allows the collected multimodal data to share a common semantic space that facilitates the prediction of individual variables. Each fused numerical vector was fed into an unsupervised clustering algorithm called Self-Organizing-Maps (SOM). The trained SOM serves as a predictor for future cases, allowing predicting consequences such as total deaths, total people affected, and total damage, and provides specific recommendations for assessments in the shelter and housing sector. To achieve such prediction, a satellite image from before the disaster and the geographic and demographic conditions are shown to the trained model, which achieved a prediction accuracy of 62 %

cs.LG↗