Main menu: Home | Resume(CV) | Projects | Teaching | Tutorials | Blog | Hobbies | Contact Me


Clinical Electronic Health Record Outlier Detection on Temporal DataIn this work we focus on developing both Machine Learning algorithms and a distributed scalable infrastructure to build real time alerting system by processing millions of time series information of patients to automatically learn and detect possible mistakes by doctors at ICU sections of UPMC hospitals. We learn models on temporal data extracted from machines, prescriptions of doctor, lab records and etc. to detect anomaly behavior in doctors behavior for a patient.

Minimal Temporal Predictive Pattern Mining considering features' causality: We have been able to show that using causal relationships between features in datasets we can drive rules to filter many patterns or prevent generating these patterns. Our early results have shown that the number of these patterns can be significantly reduced in some datasets taking causal relationships into consideration while maintaining decent classification accuracy.

Veerak is a smart home project that I started as a hobby in summer 2014 while I had gone back to my home country Iran. Basically, the idea started as I noticed that my family had been having problems with watchin movies and even internet as I had stepped out. Therefore, I decided to start a smart home project as a hobby and improve it every year. Currently, the project includes a media server, a web based file server. Read More