Areas of Research
Research Philosophy
ADIA Lab explores ideas with important, practical applications in the world. We support a wide range of research topics including underfunded subjects because we understand that important discoveries can be made in under-explored places.
Collaboration is at the core of what we do. We connect some of the world’s leading minds in data and computational science to solve challenges through interdisciplinary collaboration.
High level areas of interest
Open problems, that ADIA Lab is interested in, include:
Causal discovery and experimental design:
Almost all of science aims at identifying causal relations, and the mechanisms behind them. Causal science is a new and thriving area of research that promises to uncover insights that, until now, have remained hidden in observational studies.
AI in inverse design:
AI can help develop systems designed to have desirable properties, with important applications to materials, chemistry, biology, microelectronics, energy technologies, and other engineered systems.
AI automation and robotics:
The idea of automating research and production has been a goal of scientists for many decades. Recently progress has been made in small pilot projects from materials science, chemistry, biology, advanced imaging, and other areas, to optimize the discovery process.
Inference from noisy and/or incomplete data:
We often do not have complete information about all variables involved in a system. Yet, data scientists want to build models from such incomplete information to do prediction and inference.
High-performance computing:
High Performance Computing (HPC) is essential for making further progress in large scale ML models, and analyzing and managing large data sets. New architectures for AI, heterogenous systems with accelerators, efficient layout of massive amounts of data for fast access, new communications and network capabilities, and focus on energy efficiency will change the future of HPC. The potential of quantum computing needs to be understood.
Interpretable machine learning:
While machine learning has made important strides in recent decades, its widespread adoption has been hampered by the black-box nature of the resulting algorithms. To solve this problem, researchers must develop methods that help humans understand the nature of the pattern discovered by machines.
Digital economy, distributed ledgers and tokenization:
Digital economies will rely on digital currencies for payments, distributed ledgers for recording those transactions, and tokens for representing ownership of physical assets.
Audit, control and fraud detection:
To fully capitalize the benefits of automation industrial processes and digitization of the economy, it is important to develop systems for auditing those activities and discover when bad actors may attempt to profit from a system vulnerability.
Cybersecurity:
As more of society’s economic activity goes digital, scientists must develop the defenses needed to ensure the integrity of those systems. Leverage AI based tools to better detect threats and protect systems and data.