High-throughput computations have provided access to multiple properties of several inorganic materials. High-throughput experiments have enabled generation of materials properties under operating conditions. Both these approaches have accelerated discovery of materials. Further acceleration of materials discovery requires integration of these approaches that have different scopes and underlying assumptions.

In this talk, I will present novel approaches based on network theory to predict synthesizability and hence reduce the "synthesis gap" between theory and experiments. Another significant gap between theory and experiment is the "characterization gap" which is due to the need for laborious characterization methods to capture an experimentally synthesized material's fingerprint. In this talk, I will present the use of machine learning methods to address this gap to capture structural, chemical, and electronic fingerprints of materials.

High-throughput experiments have played a significant role in accessing high-dimensional chemical and screening space for materials discovery. However, practical implementations of high-throughput experimentation focuses on a specific region of the large chemical and screening space due to the exponential increase in number of experiments needed as dictated by the curse of dimensionality. I will present further acceleration of high-throughput experimentation by using machine learning approaches. Particularly, approaches that utilize information obtained from high-throughput computations will be discussed.