Connecting the world and beyond

PP 26

Project Resilience

​​​​​​​​​​​​​​​​​​​​​​​​​​​          Project Resilience Pink-Mauve.png aianddatacommonsbow-01.png​

​Project Resilience was initiated under the Global Initiative on AI and Data Commons to build a public AI utility where a global community of innovators and thought leaders could enhance and utilize a collection of data and AI approaches to help with better preparedness, intervention, and response to environmental, health, information, or economic threats to our communities, and contribute the general efforts towards meeting the Sustainable Development Goals (SDGs)​.
 
Project Resilience begun with predictive methods to help with health interventions to help contain COVID-19 threats, and extended this to a framework for building an AI utility to address other areas. ​The initial framework focused on climate and energy-related SDG targets, such as Target 13.2 “integrate climate change measures into national policies, strategies and planning” (under Goal ​13, Climate Action​), Target 7.2  “increase substantially the share of renewable energy in the global energy mix” (under Goal 7, Affordable and Clean Energy​). 

Project Resilience’s AI utility was intended to provide policy makers, industry, academia, and NGOs with the needed insights to act on the SDGs and meet the targets for their localized areas. The AI Utility was intended to provide a free, always on, online predictive service that will provide reliable information for decision making to determine the best path for each localized area to make progress on these economic, social, and environmental challenges. ​​​​​​

Working Groups

​1. Minimum​ Viable Product (MVP) Working Group

This Working Group was composed of subject matter experts in machine learning modeling, UX development, Architecture and DevOps who are working towards the goal of creating an MVP for the machine learning ensemble model and supporting architecture for one of the SDG topics.

The MVP working group worked in two Tracks (Data and Architecture) to produce the following deliverables:
  • Develop architecture to pull input and output data hosted by third parties
  • Develop code to compare both predictors and prescriptors in third party models and produce a set of performance metrics
  • Build a portal to visualize assessment of predictors and prescriptors to include generations of key performance indicators (KPIs) and comparison across models
  • Develop ensemble model for predictors and prescriptors
  • ​Build API for third parties to submit models​
The working group developed a first application on land-use optimization, read more ​from the arXiv paper "Discovering effective policies for land-use planning​"​.​ The paper won the "Best Pathway to Impact" Award at the NeurIPS 2023 Workshop on Tackling Climate Change with Machine Learning; see the pre-recorded talk, slides, poster, and a short version of the paper at the workshop site​, and try out the interactive demo​ of the system.​
​​
MVP Demo​​​  Land Use O​ptimization demo, AI for Good Global Summit Workshop, 5 July 2023​

​​MVP Working Group Lead​​​  Baba​k Hodjat​​, Cognizant, ​United States

​2. Data Working Group

This Working Group ​was composed of subject matter experts in data science, data sharing, and data standardization. The goal of the data working group was to provide guidelines for data contribution towards solutions that project resilience will create, and develop specifications and conduct case studies for data sharing in a standardized way with interoperable interfaces with the following specific objectives:
  • To identify contributors and their roles (relationships) as data suppliers (sources)
  • To convert collected data into publicly available data and/or datasets
  • To validate data quality with appropriate KPIs
  • To support data clearing house as a platform to aggregate the data
  • To curate data with common data models for shared taxonomy
  • To support data features (context/action/ outcomes) and repositories (local storages)
  • To support data life cycle management
  • To ensure security, privacy, and trust as well as legal compliance including data ownership
Data Working Group Lead​​​  Gyu Myoung Lee, Liverpool John Moores University, United Kingdom 

​​

​​



© ITU All Rights Reserved

Back to top