AI for Drug discovery
Drug-target interaction (DTI) prediction and protein stability change prediction upon mutation are carried out by deep learning methods. Our methods requires less time and domain expertise, while achieves higher accuracy.
.png)
CHIP MANUFACUTRING DEFECT CLASSFICATION
The wafer defect played a important role in yield improvement. We explore deep learning methods to classify the defect types for high throughput and accuracy.
Soft robotic's self proprioception
Soft robotics is hard to model. We embed strechable hydrogel flexible sensor to sense the shape change and use the data for kinematics modelling
Multi-modal learning for soft robotics
Vision and tactile sensing are crucial for robot perception. Our mission is to explore the multi-modal perception based on probabilistic statistic
a Machine Learning based Method for Differential Scanning Calorimetry Signal Analysis
we propose a method for automatically processing differential scanning calorimetry (DSC) data, which can estimate the baseline signal and interpret the net peak signal data.
Meanwhile, we used reinforcement learning to separate the peaks as sub-peaks for protein domain study.
