Updates:

• (11/13) Students interested in research opportunities can sign up through the following link.
• (10/31) Our work related to Causal Machine Learning in child placement decisions got published In proceedings of European Conference of Artificial Intelligence 2023. (ECAI’23) (link)

Research:

In a world where machine learning (ML) systems are increasingly integrated into real-world applications, the importance of ensuring responsible and trustworthy behavior is paramount. Achieving trustworthiness in ML goes beyond mere accuracy; it encompasses attributes like explainability, fairness, privacy preservation, causality, and robustness. This overarching quest for trustworthiness characterizes the field of Trustworthy Machine Learning.

My research endeavors are set to identify applications where multiple trustworthiness attributes are simultaneously valued. I aspire not only to identify these complex domains but also to explore potential solutions. An intriguing avenue lies in the realm of multi-view data, investigating its potential to enhance trustworthiness in ML. While multi-view data has been extensively used for performance optimization, there’s a noticeable gap in its application for building trustworthy ML models. Thus, my research aims to delve deeper into this promising direction, harnessing the power of multi-view data to create robust and dependable ML systems for diverse applications.

Furthermore, I am keen to explore the applicability of ML models in decision making, particularly in contexts where randomized controlled trials are challenging to implement. My goal is to automate the training and validation of models across diverse applications, making causal ML more accessible to practitioners from various fields.

Another area of interest lies in democratizing ML. The vision is to make this technology widely accessible. A critical consideration here is the hardware requirements. I’m particularly interested in the research field of “tiny ML,” which focuses on enabling smaller and simpler models to achieve performance comparable to larger, more complex models. The aim is to make ML more accessible and feasible for a broader audience.

Featured Publications:

• Sohaib Kiani, Jared Barton, Lynda Heimbach, Jon Suchinsky and Bo Luo. Counterfactual Prediction under Selective Confounding. In proceedings of European Conference of Artificial Intelligence 2023. (ECAI’23) (link)
• Sohaib Kiani, Sana Awan, Chao Lan, Fengjun Li and Bo Luo. Two Souls in an Adversarial Image: Towards Universal Adversarial Example Detection using Multi-view Inconsistency. In Proceedings of Annual Computer Security Application Conference 2021. (ACSAC’21) (Distinguished paper award) (link)
• Sohaib Kiani, Sana Awan, Jun Huan, Fengjun Li and Bo Luo. WOLF: Automated Machine Learning Workflow Management Framework for Malware Detection and other Applications. In Proceedings of the 7th Symposium on Hot Topics in the Science of Security 2020. (HotSos’ 20) (link)
• Ningqing Qian, Sohaib Kiani, Bahareh Shakibajahromi. Improved Poisson Surface Reconstruction with various Passive Visual Cues from Multiple Camera Views. In 7th Pacific Rim Symposium on Image and Video Technology, Springer, Heidelberg 2015. (PSIVT’ 15) (link)