Janani Kalyanam

Graduate Student, University of California - San Diego


Mountain View

  • PhD [in progress completed!], University of California, San Diego
    Advisor: Gert Lanckriet
  • MS, University of Wisconsin, Madison, 2009.
  • BS, Rutgers University, New Brunswick NJ, 2007.

Research I develop machine learning based algorithms to extract useful information from microblogging data, especially Twitter (which is inherently very noisy). My days are usually spent doing some combination of the following:
  • writing scripts (in Python),
  • developing algorithms (Math),
  • converting the algorithms to code (again, scientific computing in Python)
  • writing papers (LaTeX).



  • Prediction and Characterization of High-Activity Events in Social Media Triggered by Real-World News
    J. Kalyanam, M. Quezada, B. Poblete, G.R.G. Lanckriet
    [accepted] PLoS ONE, 2016
    link, cite.
  • Exploring Trends of Nonmedical use of Prescription Drugs and Polydrug Abuse in the Twittersphere Using Unsupervised Machine Learning
    J. Kalyanam, T. Katsuki, G.R.G Lankriet, T.K. Mackey
    Addictive Behaviors, 2016
    link, cite.
  • From Event Detection to Story Telling on Microblogs
    J. Kalyanam, S. Velupillai, M. Conway, G.R.G. Lanckriet
    Proceedings of the ACM/IEEE Conference on Advances in Social Network Analysis and Mining, 2016
    pdf, cite, slides, code.
  • Levaraging Social Context for Topic Evolution
    J.Kalyanam, A. Mantrach, D.Saez-Trumper, H. Vahabi, G.R.G. Lanckriet
    Proceedings of 21st ACM Conference on Knowledge Discovery and Data Mining, 2015
    pdf, cite, code, slides, poster
  • Facts and Fabrications about Ebola: A Twitter Based Study
    J.Kalyanam, S. Velupillai, S. Doan, M. Conway, G.R.G. Lanckriet
    ACM KDD Workshop on Connected Health in Big Data Era, 2015.
    pdf, cite, data, slides
  • Learning from Unstructured Multimedia Data
    J.Kalyanam, G.R.G. Lanckriet
    Proceedings of 23rd International Conference on World Wide Web, 2014
    pdf, cite
  • Identification and Compensation of Wiener-Hammerstein systems with feedback
    A.Bolstad, B.Miller, J.Goodman, J.Kalyanam
    Proceedings of 36th Internation Conference of Acoustics, Speech and Signal Processing, 2011
    pdf, cite
  • Physical Layer Considerations for Wideband Cognitive Radio
    J.Goodman, B.Miller, A.Bolstad, J.Vian, J.Kalyanam
    Military Communications Conference, 2011
    pdf, cite
  • Probabilistic Algorithm for List-Viterbi Decoding
    Janani Kalyanam
    MS Thesis, 2009.
    pdf, cite
  • Distributed Spatio-temporal Spectrum Sensing: an experimental study
    C.Raman, J.Kalyanam, I.Seskar, N.Mandayam
    Asilomar Conference on Signals, Systems, and Computers, 2007
    pdf, cite

Random Musings

So, I never thought I would be one of those people who is touchy about their terminal/editor/etc set-ups. As it turns out, I am. These are some of my CURRENT preferences.
  • I'll always UBUNTU.
  • TERMINAL: Solarized. With a bit of modification to the font color. #B0C5C8 to be precise. The font is NIMBUS MONO L BOLD.
  • TERMINAL: I don't like the entire path being shown at the prompt. I change the .bashrc to:
    if [ "$color_prompt" = yes ]; then
        #PS1='${debian_chroot:+($debian_chroot)}\u@\h:\w\$ '
        PS1='${debian_chroot:+($debian_chroot)}\u@\h:\W\$ '
  • EDITOR: I'll always VIM. I am right now with railcasts (sometimes, I do solarized-dark). Clone the repos for both, and place the *.vim in ~/.vim/colors/. Then modify your ~/.vimrc to set the colorscheme. NOTE: Solarized has a light and dark versions.
  • EDITOR: I always set tabstop=4
  • FONT: So, VIM takes over the font from the terminal. On a laptop, I prefer the default font for the terminal and vim. On larger screens, I prefer NIMBUS MONO L BOLD.
  • Some other stuff - to clear stuff at boot:
    uname -r # lists your current kernel.  
    dpkg --list | grep linux-image # grabs all the images past and current.
    sudo apt-get purge linux-image-$.$.$.$. # get rid of the old images
    Then, update grub and reboot.