About UN Global Pulse
UN Global Pulse works through a network of labs to accelerate the discovery, development, and responsible use of big data and artificial intelligence innovations and policies for sustainable development, humanitarian action, and peace.Activities by UN Global Pulse
A computational framework for predictive modeling of refugee and IDP movements
Global UgandaBrazilMaliBurundiCentral African RepublicSomaliaAfghanistanUkraineBangladeshThis project provides a systematic model-agnostic framework for structuring the prediction problem of refugee and IDP movements. It aims to facilitate a more comprehensive understanding of this emerging field of research for humanitarian aid agencies.
Activity Type Policy/Regulatory GuidanceResearch/Reports/AssessmentsOnline radio monitoring for public health social listening and beyond
AfricaA radio monitoring pipeline was developed to 'listen' to online radio stations, transcribe the audio using machine learning speech-to-text models, and analyze the content using NLP methods. The dashboard is used by infodemic managers and decision-makers to inform public health interventions.
Activity Type AI Tools/SolutionsResearch/Reports/AssessmentsCoordiMap: From Crisis to Solution with Coordinated Shelter & Settlements Responses
Global El SalvadorSyriaCoordiMap is an innovation initiative designed by the Global Shelter Cluster (GSC), under the data innovation fund provided by UNHCR Innovation Service, to improve coordination and decision-making across the humanitarian response lifecycle. The project combines artificial intelligence (AI), machine learning (ML), satellite imagery, geographic information systems (GIS), and participatory assessment methods to deliver a coordinated framework for shelter and settlement responses.
Activity Type AI Tools/SolutionsResearch/Reports/AssessmentsPulseSatellite: A collaboration tool using human-AI interaction to analyse satellite imagery
Africa, Asia and the Pacific BangladeshSomalia
PulseSatellite is a collaborative satellite image analysis tool that leverages neural network models that can be retrained on-the-fly and adapted to specific humanitarian contexts and geographies. The tool has models for mapping structures in refugee settlements, roof density detection, and flood mapping.
We are also working closely with UNOSAT to develop benchmark datasets for shelter (refugee camp) mapping, building footprint detection and damage assessment. We plan to use these to test many of the available well trained and top-performing models, but in the context of UN-focused datasets (e.g. with more of a Global South and development context than many of the standard machine learning benchmarks) and make this available as a service to the UN system.
Operational contexts are rapidly changing, meaning that AI models may not always perform well. Through using a human-in-the-loop approach we have found that models can be adapted to such changing settings, however, this still requires (sometimes significant) manual intervention from analysts.