AI-based Discovery and Innovation

Using artificial intelligence (AI)-based approaches (especially manifold learning) for understanding the delicate rules of nature and utilizing them for forming innovative software- and hardware-based tools for addressing the major challenges in a wide range of disciplines.

Innovative ideas in science and engineering are usually formed by understanding the delicate relations between different phenomena or constituents of large systems and utilizing this knowledge to address the existing challenges. Such innovations in many cases are formed by exhaustive searches, trial and errors, or sometimes by luck. Some rules are trivial, and some are discovered by researching a subject for decades. Recent surges in availability of data and processing resources in a wide range of disciplines have inspired new research, under the umbrella of AI, in solving the major challenges in science and engineering. Nevertheless, most of the presented techniques have been focused on problem-solving (e.g., optimization, design, and analysis). The main goal of this VIP team is to focus on using novel artificial intelligence approaches (especially, manifold-learning techniques) in forming systematic approaches to learn the unidentified governing rules and axioms in different disciplines (e.g., physics, biomedical science, psychology, art, and finance) and to utilize them to develop new software- and hardware-based tools with transformative impact in a wide range of applications from engineered nanostructures for medical imaging and sensing, brain probing, optical computing, and quantum information processing to intelligent data-processing software for detecting the delicate patterns in blood biomarkers, financial markets, and social media. As such, the research activities are defined in two major thrusts that are focused on 1) designing new devices and systems that utilize nanotechnology to solve engineering challenges in a wide range of disciplines and 2) forming intelligent learning algorithms for extracting extremely valuable knowledge from large sets of data (e.g., images, spectral information, dynamics of variations, etc.).