I am a postdoctoral researcher in the Probabilistic Machine Learning (PML) group at the Department of Computer Science, Aalto University, Finland, under the supervision of Prof. Samuel Kaski. My research focuses on personalized medicine, specifically building machine learning methods for medical networks and graphical data. In particular, I am working on developing causal inference and robust graph neural network (GNN) models.

Background:

Earlier I was a Postdoctoral fellow at Indian Statistical Institute (ISI) Delhi. Here I was working under Prof. Isha Dewan and Dr. Sudheesh Kumar Kattumannil (from ISI Chennai). During this position, I completed two research papers with them. My work focused on developing and applying nonparametric approaches for constructing confidence intervals for the volume under ROC surfaces, as well as deriving estimators for a new extropy measure of divergence. In particular, we generalized the jackknife empirical likelihood (JEL) method to more complex and multivariate data settings. Additionally, we derived estimators using traditional methods such as U-statistics, kernel-based methods, and empirical approaches, and demonstrated various real-world applications of the proposed methods.

Prior to this, I completed my Ph.D. from the Department of Mathematics and Statistics at the Indian Institute of Technology (IIT) Kanpur, under the supervision of Prof. Neeraj Misra. My Ph.D. work is centered around "the Estimation of order-restricted parameters." The problem of estimating the location (scale) parameters of two distributions when the ordering between them is known apriori has been extensively studied in the literature. Many of these studies are centered around deriving estimators that dominate the best location (scale) equivariant estimators or the maximum likelihood estimators (MLEs) by exploiting the prior information that parameters are order restricted. In my thesis, I attempt to unify various results in the literature by considering a general bivariate location (scale) probability model and a general loss function. Some novel results in this area are included in my research.

My research interests include

Machine Learning for Medical and Graph-Structured Data
Causal inference, robust graph neural networks, personalized medicine.

Nonparametric and Empirical Likelihood Methods
ROC surfaces, survival/extropy-based divergence, U-statistics, confidence regions.

Order-Restricted Statistical Inference
Isotonic regression, shrinkage estimators, generalized Pitman nearness, decision-theoretic estimation.

The presentation file for my talk (2024) at Indian Statistical Institute (ISI) Delhi, titled "The Receiver Operating Characteristic Curve, Its Applications, and Extensions" is available here .

Contact

Email: garg.naresh22@gmail.com, naresh.garg@aalto.fi

Address: Computer Science Department, Aalto University, Finland (02150)