Perpetual Abstraction     About     Research     Resume     Archive     Feed
Ramblings of a rogue Mathematician

Research

Interests

Uncertainty

  • Bayesian Inference
  • Stochastic processes
  • Uncertainty Quantification

Applied Mathematics

  • Functional Analysis
  • Physical Systems, Mathematical Physics
  • Scientific Computing

Machine Learning & Artificial Intelligence

  • Kernel based machine learning models: Gaussian Processes, LS-SVM
  • Time series/autoregressive models.
  • Machine learning software

Publications

2017

  • Probabilistic Forecasting of the Disturbance Storm Time Index: An Autoregressive Gaussian Process approach: M. Chandorkar, E. Camporeale, S. Wing.

    Space Weather, American Geophysical Society. (accepted)

2016

  • On the propagation of uncertainties in radiation belt simulations: E. Camporeale , Y. Shprits , M. Chandorkar, A. Drozdov, S. Wing

    Space Weather, American Geophysical Society.

2015

  • Fixed-Size Least Squares Support Vector Machines: Scala Implementation for Large Scale Classification: M. Chandorkar, R. Mall, O. Lauwers, J. A. K Suykens, B. De Moor.

    IEEE Symposium Series on Computational Intelligence: IEEE Symposium on Computational Intelligence in Big Data (2015 IEEE CIBD). link


Projects

PhD

Sept 2015 -

Multiscale Dynamics, CWI Amsterdam & TAO INRIA Paris-Saclay

  • Title: Machine Learning for Space Weather (link)
  • Supervisor: Enrico Camporeale & Michele Sebag
  • Description: Applying Machine Learning and Statistical Inference techniques to problems of Space, Magnetospheric and Heliospheric Physics. This includes:
    1. Formulating, implementing and testing data-driven prediction models for geomagnetic activity
    2. Performing statistical inference of important Magnetospheric parameters using satellite measurements.
    3. Using variety of data sources to inform and improve physical understanding and prediction capability of geomagnetic phenomena.

Masters Thesis

Oct 2015 - Sept 2016

ESAT-STADIUS, KU Leuven

  • Title: FS-LSSVM: A Scala based programming framework for Large Scale Classification (link)
  • Supervisors: Johan Suykens, Bart De Moor
  • Description: A implementation of the Fixed Size Least Squares Support Vector Machines in the Scala programming language, using Apache Spark RDDs.

Open Source


  • DynaML: DynaML is an open source library targeted towards machine learning researchers and practitioners, written in the Scala programming language.

  • PlasmaML: A multi-language open source toolbox for machine learning applications in space plasmas.

  • ScalaImageToolbox: Scala language implementation of Active Shape Models for the purpose of identification of incisors in dental radiographs. Submitted as a class project for the course Computer Vision. See the report for more details.

  • recsys2014: This repository is a recommender algorithm evaluator based on the RecSys Challenge 2014. It uses the Twitter data set and generates recommendations as well as ranks tweets from the test set based on their predicted ‘engagement’ count.