Current Mentoring

Mentoring is key to developing the next great thinkers. Thus, we are actively looking to engage and support upcoming researchers and groups. On his own time, Dr. Yasin actively works with students and groups to expand their research interests beyond the scope of theory. He is currently working with many researchers in various areas. He participates in independent study, where he mentors students looking to do advanced work in a specific topic area. Students participating in the independent study attain publications from the resulting work and learn various skills needed to expand their future careers. Listed below are some research areas that he is currently working and mentoring in.

One of his students is working on developing early cancer detection algorithms. For example, early-stage tumors from X-ray images, planning to research the difference in blood samples between CRC patients and healthy people, and then developing procedures for labeling biomarkers useful for CRC diagnosis.

Another student is working on DNA/RNA methylation, and we are interested in sequencing, because the occurrence of this methylation displays strong sequence bias, and it can serve as a biomarker to indicate different biological states. It’s hard to interpret what features have been learned by the models if we only feed them with sequences. Thus, we will develop models to better provide insight into the study of the biological mechanism.

Another project pertains to AI classification models in cancer imaging datasets. We aim to do image classification using deep CNNs on X-ray images.

Another student is working with him on dynamically evolving AI and ML models that have no set hyperparameters to define a static structure but to evolve the structure dynamically. This work aims to better understand the topology considerations when defining a neural network.

Some other interesting areas we are looking at are continuous learning models, explainable ML/AI, developing and evaluation networks with corruption architecture and characterizing their robustness.

Most recently during his independent study class, one of his students was working on transcriptomics data set. We explored various methodologies and frameworks towards effective representational learning of spatial transcriptomics data, and by extension, develop an optimized pipeline allowing for real-time analysis of this data. In the upcoming months we will publish our findings and look forward to continuing this work.

He also actively mentors’ students looking to teach in the near future. Those students support him in teaching assistance roles, this allows exposure to academic teaching at a university level. If you are interested in some of this work or the derivatives, please feel free to reach out. We are actively looking to support and mentor any curious mind.