http://arun.chagantys.org/ 2012-02-22T22:27:01+03:00 http://arun.chagantys.org/ http://arun.chagantys.org/lib/tpl/apricot/images/favicon.ico text/html 2012-02-03T14:08:39+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/wiki:sidebar?rev=1328267319&do=diff * Home * blog * Research * Projects * Teaching * Links * About text/html 2011-12-21T14:46:30+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/home?rev=1324467990&do=diff My name is Arun Tejasvi Chaganty. I am currently in my fifth year of a dual degree (BTech. + MTech.) in Computer Science and Engineering at the Indian Institute of Technology, Madras . I'm passionate about the machine learning and optimisation, and have some experience. text/html 2011-12-21T14:44:35+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/research:bayez?rev=1324467875&do=diff BayeZ (With: Aditya Nori, Sriram Rajamani and Akash Lal) Abstract Statistical relational learning consists of inferring new relations from an existing data corpus using probabilistic formulas as specifications. We advance the state-of-the-art in relational learning significantly using techniques developed in program verification. In particular, our technique efficiently deals with axioms which are probabilistic formulas with probability 0 or 1. These axioms are required to be satisfied by any… text/html 2011-12-21T14:42:42+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/research:small-world-rl?rev=1324467762&do=diff Learning in a Small World (With: Balaraman Ravindran, Prateek Gaur) Abstract Understanding how we are able to perform a diverse set of complex tasks has been a central question for the Artificial Intelligence community. One popular approach is to use temporal abstraction as a framework to capture the notion of subtasks. However, this transfers the problem to finding the right subtasks, which is still an open problem. Existing approaches for subtask generation require too much knowledge of the… text/html 2011-12-21T14:41:54+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=research%3Aaamas-2012-submitted.pdf&ns=research&rev=1324467714&tab_details=history&mediado=diff&do=media text/html 2011-12-21T14:39:41+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/research:home?rev=1324467581&do=diff I am primarily interested in machine learning and its applications. Current Work Homomorphisms in Reinforcement Learning across Continuity and Partial Observability (2011 - Present) (With: Balaraman Ravindran) Short abstract: Markov Decision Process (MDP) homomorphisms map states and actions from one MDP to another, providing a basis for the transfer of learning, and MDP minimisation. Homomorphisms in the discrete MDP setting have been comprehensively studied; only recently has work been d… text/html 2011-12-12T14:56:11+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=research%3Asiads.pdf&ns=research&rev=1323690971&tab_details=history&mediado=diff&do=media text/html 2011-12-07T20:53:26+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=me%3Aresume.pdf&ns=me&rev=1323280406&tab_details=history&mediado=diff&do=media text/html 2011-12-07T20:38:13+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/research:collapsed-ctm?rev=1323279493&do=diff Collapsed Inference Methods for Non-conjugate Correlated Topic Model (2011 - Present) (Mentor: Balaraman Ravindran) Abstract The Correlated Topic Model (CTM) was proposed by Blei et. al to capture topic correlations. It does so by assuming the topics are drawn from a log-normal distribution. The log-normal and multinomial distributions are not conjugate, thus leading to complex inference and learning algorithms. We outline several approximations we explored to `collapse' the CTM to a simpler … text/html 2011-12-07T20:38:01+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=research%3Actm-cvb-report-dec-2011.pdf&ns=research&rev=1323279481&tab_details=history&mediado=diff&do=media text/html 2011-12-07T20:35:03+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/research:rl-homomorphisms?rev=1323279303&do=diff Homomorphisms in Reinforcement Learning across Continuity and Partial Observability (2011 - Present) (Mentor: Balaraman Ravindran) Abstract Markov Decision Process (MDP) homomorphisms map states and actions from one MDP to another, providing a basis for the transfer of learning, and MDP minimisation. Homomorphisms in the discrete MDP setting have been comprehensively studied; only recently has work been done to define homomorphisms in continuous or partially observable domains. The objective … text/html 2011-12-07T20:34:48+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=research%3Arl-hom-thesis-proposal.pdf&ns=research&rev=1323279288&tab_details=history&mediado=diff&do=media text/html 2011-12-07T20:27:23+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/research:dynamical-population-models?rev=1323278843&do=diff Some models in population dynamics: Stability, limit cycles, chaos and implications on control (With: Gaurav Raina, Ramya Korlakai Vinayak) Abstract Biological systems are ripe with phenomenon occurring across various timescales, yet conventional system design principles steer us away from the very phenomena that make biology interesting. It is not obvious that conventional design wisdom applies in this setting. We shed some light on the topic through the study of two of the simplest quadrati… text/html 2011-12-07T20:26:20+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=research%3Adynamical-population-models-milestone4.pdf&ns=research&rev=1323278780&tab_details=history&mediado=diff&do=media text/html 2011-12-07T20:26:07+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=research%3Adynamical-population-model-milestone4.pdf&ns=research&rev=1323278767&tab_details=history&mediado=diff&do=media text/html 2011-12-07T16:37:19+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/projects:vimjuta?rev=1323265039&do=diff Vimjuta - Vim + Anjuta = World Domination Vimjuta (or more formally anjuta-gvim) is an Anjuta Plugin which integrates Vim into the IDE. Finally get to use a debugger and a project manager in conjuction with your 1337 vimming skills, to move to the ultimate goal. World Domination. text/html 2011-12-07T16:28:17+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/projects:home?rev=1323264497&do=diff Projects A majority of my projects are on github. Working * Webbed Feet : Online AI game competition application * R0botnik : Extensible python-based IRC bot * pp-editor : An experiment to see if we can simultaneously read a story from two perspectives. text/html 2011-12-07T15:57:17+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/projects:on-the-fly?rev=1323262637&do=diff OnTheFly - Tomboy Addin Supports on the fly wikitization. Note OnTheFly has been pwned by this plugin . Example + Features Supports the following formatting tags: * *bold* = bold * /italic/ = italic * +strikethrough+ = strikethrough * =highlight= = text/html 2011-12-07T15:48:04+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/?image=projects%3Aonthefly-0.1.tar.gz&ns=projects&rev=1323262084&tab_details=history&mediado=diff&do=media text/html 2011-12-07T15:35:11+03:00 Arun Tejasvi Chaganty http://arun.chagantys.org/projects:robotnik?rev=1323261311&do=diff Rob0tnik - Taking over the world, one IRC channel at a time Rob0tnik is an extensible IRC bot, a key ingredient to anyone's world domination plan. Based on twisted's IRC protocol, and featuring an clean plugin system, modelled along the lines of Django, Rob0tnik should be easy to extend (I hope).