Natalia Hernandez Gardiol
MIT CSAIL, 32-G585
32 Vassar Street
Cambridge, MA 02139
nhg at csail.mit.edu
   

My cv.
I am currently located in the San Francisco Bay Area, where I work on search and recommendation algorithms, text modeling, big data, and other machine-learning related projects.
From March - July 2010, I was a postdoc/visiting professor at U Pompeu Fabra, working with Hector Geffner.
Until February 2010, I worked at State Street Associates in Cambridge, MA.
In Spring 2008, I was a postdoctoral researcher in CSAIL (the former AI Lab) at MIT.

I finished my PhD at MIT in December 2007, where I worked with Leslie Kaelbling in the LIS lab on planning in stochastic, relational domains.
In 2003, I received my MS from MIT. I did my undergraduate work at Michigan State University, where I worked with Sridhar Mahadevan on hierarchical reinforcement learning.

Papers.

  • Natalia H. Gardiol and Leslie Pack Kaelbling, "Adaptive Envelope MDPs for Relational Equivalence-based Planning", MIT Technical Report MIT-CSAIL-TR-2008-050, July 2008. [pdf][bib]

  • Natalia H. Gardiol, "Relational Envelope-based Planning", Ph.D. Thesis, MIT CSAIL, December 2007. [pdf][bib]

  • Natalia H. Gardiol and Leslie Pack Kaelbling, "Action-space partitioning for planning", AAAI 2007, July 2007. [pdf][bib]

  • Natalia H. Gardiol and Leslie Pack Kaelbling, "Computing action equivalences for planning", ICAPS 2006 Doctoral Consortium, June 2006. [pdf] [bib]

  • Natalia H. Gardiol and Leslie Pack Kaelbling, "Computing action equivalences for planning under time-constraints", MIT Technical Report MIT-CSAIL-TR-2006-022, December 2005. [pdf][bib]

  • Natalia H. Gardiol and Leslie Pack Kaelbling, "Envelope-based Planning in Relational MDPs", Advances in Neural Information Processing Systems 16 (NIPS*2003) [pdf][bib]

  • Natalia H. Gardiol, "Applying Probabilistic Rules To Relational Worlds", M.S. Thesis, MIT AI Lab, November 2002. [ps.gz][pdf][bib]

  • Sarah Finney, Natalia H. Gardiol, Leslie Pack Kaelbling, Tim Oates, "The Thing That We Tried Didn't Work Very Well: Deictic Representation in Reinforcement Learning", 18th International Conference on Uncertainty in Artificial Intelligence, Edmonton, August 2002 (UAI-02). [ps.gz][pdf][bib]

  • Sarah Finney, Natalia H. Gardiol, Leslie Pack Kaelbling, Tim Oates, "Learning with Deictic Representation", MIT AI Lab Technical Report (AIM-2002-006), April 2002. [ps.gz][pdf][bib]

  • Leslie Pack Kaelbling, Tim Oates, Natalia Hernandez Gardiol, Sarah Finney, "Learning in Worlds with Objects", Working Notes of the AAAI Stanford Spring Symposium on Learning Grounded Representations, March 2001. [ps.gz][pdf][bib]

  • Natalia Hernandez Gardiol and Sridhar Mahadevan, "Hierarchical Memory-based Reinforcement Learning", Advances in Neural Information Processing Systems 13, (NIPS*2000). MIT Press, Cambridge, 2001. [ps.gz][pdf][bib]
    An earlier draft.


    .cycling
    .mitoc
    .banff, 2002
    .hockey pix (old)

  • Click for Cambridge, Massachusetts Forecast
    (SMOR) <--- (TNFB)