Overview
Projects
Computers
Research Stories
Collaboration
Software & Servers
Our research focuses on development of deep learning, machine learning, data mining, and evolutionary algorithms for scientific discovery and engineering innovation in materials science, bioinformatics, health informatics, function genomics, medical and health sciences, engineering designs, intelligent manufacturing and etc. We are specialized in developing state-of-art image recognition, computer vision, pattern recognition, classification, regression, predictive modeling, inverse design using state-of-the-art models such as deep neural network models, generative adversarial networks, graph neural networks, genetic programming, symbolic regression.

Student success via team work is one of the main goals in our lab. We welcome students with strong interests in solving challenging real-world biological and materials science problems using latest deep learning and artificial intelligence techniques. You can contact us at: jianjunh at cse.sc.edu
Funding Source

-------------------------
NSF



Ph.D. student
In Machine Learning and Evolution Lab, I have been developing deep neural networks for audio based remaining tolerable strength prediction for nuclear reactor composite materials. I have also worked on developing 3D convolutional neural network with voxel grid representation. Currently, I am focusing on graph neural networks which is supercool. Applications include both materials property prediction and protein-ligand binding. Our Lab is active and our weekly group meeting is cool too.
NSF



Ph.D. student
I just joined the lab last Fall. As a starting student,I worked on applying basic machine learning (Random forest and MLP neural networks) for noncentrosymetric materials prediction problem. The knowlege requirement on the materials science part is not too much as the problem was formulated by my adivsor and our materials collaborator. I found there are many challenging but interesting problems to solve in materials discovery. I have just finished my first paper and feel proud of it considering its potential use by real materials scientists.
NSF



Ph.D.
I have just graduated in Fall 2019 and now working in Facebook as research scientist. My daily work is mostly on application and design of deep learning algorithms and especially on all kinds of embedding. My training experience at USC MLEG is great. Although I worked on several bioinformatics problems, most of them are involved with deep learning algorithms, which is what the market wants. I mainly used pytorch as the deep learning framework. My experience is that it does not matter what problems you work on, the key is to learn the key principles of deep learning and how to tailor it to the problem.
NSF



Ph.D. student
I am now working on deep learning based prediction models for new materials discovery mainly dealing with chemical formula and crystal structures. We usually work on big data sets with 100,000 to 1,000,000 materials. The complexity of crystal structures and their bonds makes is much more challenging that the image data for pattern discovery or generation. I have published one paper on composition based prediction of their space group and crystal system and are ready to submit another one which studies how to predict macro materials property such as bulk modulus using electronic charge density.