Data Driven Discovery and Design (4D)

Data generated in the fields of cheminformatics, bioinformatics and materials design can be massive or sparse, time-varying or stead, low-dimensional or high-dimensional, discrete or continuous, low-fidelity or high-fidelity, uncertain and noisy, biased and typically constrained by physical laws. Such data heterogeneity and variety present a huge challenge in the adoption of existing statistical inference and prediction methods for learning models, in the interplay between experiments and computation, and in the subsequent use of these models in process-level design and optimization. In addition, the design/exploration spaces can be vast and the descriptors needed to search these spaces are usually not known a-priori. Finally, discovery and processing have been typically addressed sequentially. Yet, the ultimate process where a new material will be used provides additional constraints for the design problem, which when accounted for, can accelerate discovery and help avoid infeasible solutions. 

The 4D program aims to address the above challenges and accelerate the application of data science in industry. It provides core expertise in data analysis and data-driven discovery and design of materials and chemicals, along with collaborative opportunities across applications in chemical and biochemical engineering and materials science. Problems of interest include data sharing and management; mechanism inference and design of complex catalytic systems; adsorbent and membrane design; discovery, characterization, and design of hard materials, nanomaterials, and biomaterials; (bio) manufacturing; protein and cell engineering and manufacturing; hybrid modeling; data-driven process optimization and control. 

Contact: 

To learn more about the Data Driven Discovery & Design Program research, contact Prodromos Daoutidis. To learn more about IPRIME, contact Chris Ellison.