Artificial Intelligence for Satellite Imagery (AI4SAT)
QCAI has a number of projects using satellite imagery analysis for different tasks such as road network inference, crash risk maps, internal displacement monitoring, flood extent and vulnerability mapping, among others. This project aims to combine these individual projects in a unified framework and build customized solutions for potential end users. To this end, we […]
Monitoring Attacks on Education on Social Media
Attacks on education, students, teachers and schools have intensified in recent years. Traditional methods miss many instances of education insecurities. The Education Above All (EAA) Foundation aims to develop a Global Data Service to host data on violations against the right to education and attacks on education. We aim to assist EAA with incidents reported […]
Landslide Detection through Social Media Image Streams
Landslides occur all around the world and cause thousands of deaths and billions of dollars in infrastructural damage worldwide every year. Satellite-based landslide detection introduces data latency ranging from several hours to days. We use social sensing imagery data from Twitter to identify landslide reports in real-time. Together with our stakeholders (BGS and EMSC), we […]
Disaster Object Detection and Damage Assessment
In the aftermath of a large-scale disaster, it is important to assess the impacted area to identify damaged infrastructure such as roads, buildings, power lines. Social media can play an important role in current-day disaster management. Images shared from the disaster areas may include objects relevant to various response operations. If these objects are identified […]
Flood Extent Mapping and Vulnerability Assessment
Rapid flood extent mapping is an important task for planning response and relief operations. Moreover, vulnerability assessment of a geographical area or a population segment helps humanitarian organizations take timely and informed decisions. In this consortium project funded by QNRF, we aim to leverage two types of non-traditional data sources i.e., satellite and social media […]
Multi-omics analysis of cardiovascular diseases
The aim of the project is to study cardiovascular diseases using multi-omics data. Methodology involves simple statistical models to study each omics component, and then more sophisticated ML models will be used for combining all omics data.
Systems Research for Data-intensive Applications
This project conducts research on optimizing the storage hierarchy for data-intensive applications, focusing on (1) memory cache redesign and (2) cloud-native relational database with persistent memory disaggregation. The research is done jointly with collaborators at UTSC (China), Alibaba, and MSRA.
Efficient and Scalable Parallel Traffic Simulation
This project builds on the QarSUMO prototype that performs parallel traffic simulation, as a computation-intensive component of ML-based traffic management. The focus for the next year includes fidelity improvement (to sequential version), dynamic rerouting, and traffic-adaptive road network repartitioning.
AI and drug repurposing for diseases induced by viruses
Artificial intelligence-based compound-viral proteins interaction framework, will be used via a transfer learning setting to prioritize a set of FDA approved and investigational drugs which can potentially inhibit a specific viral proteins. To filter and narrow down the lead compounds from curated collections of pharmaceutical compounds, we use a rigorous computational framework that included homology […]
Automatic Diagnosis of 12-Lead ECG on Cardiac Arrhythmias with novel Deep Learning Neural Network model
The precision of current models restricts the use of automatic electrocardiogram (ECG) analysis in clinical practice. There is a high hope for how ML+AI technology, might enhance clinical practice. Here, we build a Neural Network-based model that will be developed using MIT Benchmark data and others to study a newly generated data from Pittsburgh hospital […]