Ongoing Research Projects
Our lab is working on forward-looking, high-impact projects in the field of AI, ML, and real-world data applications.
🔄 Concept Drift in Network Traffic
We study the phenomenon where machine learning models lose accuracy over time in dynamic environments and propose methods to detect and adapt to drift, especially in encrypted network data.
🚫 Out-of-Distribution Detection
We are developing algorithms to make models aware of and resilient to inputs that do not belong to the training data distribution—improving model reliability and safety.
🧬 Explainable ML in Healthcare
Our models for health and bioinformatics data aim not only for high accuracy, but also explainability—supporting trust, transparency, and data-driven decision making for medical professionals.
👁️ Facial Recognition Attendance
We're building a privacy-aware, real-time facial recognition system for classroom attendance to automate and enhance learning analytics for instructors.
Previous Work & Research Interests
Our lab has made notable contributions across a wide array of applied machine learning domains.
🌐 Network Traffic Analysis
Our work shows how encrypted traffic can still leak information. ISPs can identify YouTube videos even under VPN and HTTPS encryption. See: [Kha22], [AfB22].
💬 Social Media Analysis
We proposed a CNN-based framework called "HateClassify" for hate speech and offensive content detection [KhA21]. To detect genuine experts on Twitter, we introduced a modified HITS algorithm: [KhA18].
📲 IoT & Human Activity Recognition
We developed CNN-based techniques and energy-efficient feature reduction for recognizing human activity from wearable sensors: [KhA18b], [KhA18c]. Dr. Khan also co-edited the book Big Data-enabled IoT [KhK19].
Through this multi-faceted research portfolio, DSML Lab continues to contribute innovative solutions to complex problems in AI, data science, and human-centered computing.