Leveraging IEEE DataPort for Humanitarian AI

Presented at Chandigarh University in collaboration with IEEE and IEEE DataPort, April 2026.

Humanitarian crises affect 245 million people worldwide, and AI has the potential to accelerate response, but only if the data layer is solid. This talk explores how IEEE DataPort can serve as the foundation for building secure, data-driven humanitarian AI solutions.

Topics covered:
- Why humanitarian AI matters now — scale of need (UNICEF, UN OCHA, WMO data)
- Why AI fails when the data layer is weak — 32% of crisis data is outdated or unavailable
- IEEE DataPort as a platform for citable, reproducible, governed research datasets
- Reference architecture for humanitarian AI pipelines aligned with NIST AI Risk Management Framework
- Real-world use cases: disaster damage mapping and flood detection using IEEE DataPort datasets
- FAIR principles (Findable, Accessible, Interoperable, Reusable) for publishing humanitarian data
- Privacy as a non-negotiable — UNHCR data protection frameworks and NIST Privacy Framework

Transcript

1. Leveraging IEEE DataPort for Humanitarian AI – Building Secure, Data – Driven Solutions Alok Tibrewala April 05, 2026 Chandigarh University, India
2. 👋, I am Alok Tibrewala 🎓 Education MS Computer Science, The George Washington University, Washington DC 🧑💻 Work Predominantly, E-Commerce and Finance 🧠 AI Research & Architecture Hybrid AI systems | Research Papers and Book author ⚡ IEEE Leadership Senior Member | DataPort Review Board and Data Competitions SubCommittee| AI and Workforce Development Initiatives
3. What is IEEE DataPort?
4. IEEE DataPort ieee- dataport.org global repository of research dataset 20 million users 11000+ dataset
5. ACCESS 11000+ DATASETS FREE DOI FOR CITATIONS UPTO 2 TB OF STORAGE SUPPORTS MULTIPLE FILE FORMATS LINK DATASET TO PAPERS AI VISUALIZATION AND CHATBOT – COMING IN 2026 IEEE DataPort - Benefits
6. Why Humanitarian AI matters now? 245.4 million people in need 139.9 million people targeted $33.19 billion needed source: humanitarianaction.info, UN OCHA
7. Children are at the center of the need source: unicef.org, UNICEF 2026 Humanitarian Action According to UNICEF, in 2026, • 200 million children will require humanitarian assistance • requires $7.66 billion to reach 73 million children • focus areas include, • nutrition • education • health • water • sanitation UNICEF 2025 Achievements
8. Climate change is increasing the risk source: wmo.int, WMO State of the Global Climate 2024 2024 was warmest year, global mean near surface temperature about 1.55 C above 1850-1900 average extreme weather in 2024 lead to highest new annual displacement since 2008
9. Humanitarian AI fails if data layer is weak source: UN OCHA Centre for Humanitarian Data, The State of Open Humanitarian Data 2026 • 68% of crisis data was available and upto date in 2025 • 21% was available but outdated • 11% was unavailable • Metadata, governance and access control matter • Data engineering is the backbone
10. IEEE DataPort for Humanitarian AI • Responsible data before intelligent systems • Safe, ethical, effective data management is essential for humanitarian response • Access, security, governance and trust determines where humanitarian AI is usable in the field Source: UN OCHA Centre for Humanitarian Data, OCHA Data Responsibility Guidelines, revised Jan 2025
11. Dataset for Humanitarian AI Citable Robust metadata Reproducible Reputation Backed by research
12. Reference Architecture • Data source → ingestion → validation → metadata/provenance → access controls → model training → deployment → monitoring • Governance and security in the middle, not an afterthought source: nist.gov, NIST AI Risk Management Framework
13. Use case 1: Disaster damage mapping IEEE DataPort Dataset - Post Hurricane Remotely Sensed Imagery DOI - 10.21227/1s3n-f891 How it can be used? • Compare before-and-after images to identify damaged buildings • Evaluate how accurately different methods detect and outline damage • Analyze damage patterns across neighborhoods to support response and recovery planning • Create repeatable damage-assessment workflows using labeled imagery
14. Use case 2: Flood Awareness IEEE DataPort Dataset - SEN12-FLOOD : a SAR and Multispectral Dataset for Flood Detection DOI - 10.21227/w6xz-s898 How it can be used? • Map flooded areas from satellite imagery • Compare radar and optical images to improve flood detection • Track how flood conditions change over time for better situational awareness • Evaluate how well different methods classify and outline flooded areas
15. FAIR principles for publishing data • FAIR principles, • Findable • Accessible • Interoperable • Reusable • Bad metadata slows down discovery of dataset • Every dataset should ship with, • Schema • Version • Readme • Abstract source: NIH ODSS Fair Data Strategy
16. Privacy is not optional • UNHCR has formal framework for collection, use and sharing of personal data • NIST treats privacy risk across full data lifecycle • Operational baseline, • Collect less • Restrict access • Log use • Define retention source: NIST Privacy Framework
17. IEEE DataPort Demo
18. Thank you!