Research, Passion, and Innovation [News]

Welcome to UofSC Machine Learning and Evolution Laboratory (MLEG) at the Department of Computer Science and Engineering, University of South Carolina. Our research focuses on development of deep learning, machine learning, data mining, and evolutionary algorithms for knowledge discovery and innovation in material science, bioinformatics, medical and health sciences, engineering designs, and etc. We have worked on DNA regulatory motif discovery, microarray analysis, phenotype prediction/computational disease diagnosis, and gene/protein function prediction, X-ray data based phase mapping, protein-DNA, protein-ligand binding, breast cancer diagnosis based on image processing, cell segmentation, DFT based material doping, and etc.
Currently, we are working on deep learning based generative inverse design of materials and proteins, drug design,crystal structure prediction, deep learning and its applications in materials discovery, computer vision, Natural Language Processing, Audio/Sound Pattern Recognition, and fault diagnosis. We seek to develop and apply the latest artificial intelligence algorithms such as machine learning, deep learning, genetic algorithms, genetic programming, dimension reduction, non-linear mapping, sparse coding, and etc to solve challenging scientific and real-world problems.

Keywords: Machine keywords: Deep learning, Machine learning, Material Informatics, Bioinformatics, Data mining, Health Informatics;

Latest News

2022.9. Our work on deep learning based generative design of semiconductors is published in Nature npj computational materials Link. Dilanga is the lead author.
2022.5. Our scalable deep graph neural networks for materials property prediction published in Cell Patterns. We combine global attention with skip connection and group normalization to achieve state-of-the-art performance. Sadman is the lead author.
2022.3. Our MaterialsAtlas.org paper is now published in Nature npj Computational Materials. It includes more than 20 different web services for materials discovery.
2021.12. Our deep learning based crystal structure generator is the first of its kind for generating novel cubic materials. This works is published on Advanced Science . The lead author is Yong Zhao.
2022.5. Our work on applying deep transformer language model for discovery of novel inorganic materials composition is just out with manuscript released on arxiv. The accompanion web server for materials tinkering/doping is here BLMTinker
More News and Stories Here