|Aarati Parmar Martino
Department of Computer Science
Gates Building 2A -- Room 218
353 Serra Mall
Stanford, California 94305-9020
I am a graduate of the Computer Science Department PhD program here at Stanford University. I am a member of the Formal Reasoning Group, headed by Professor John McCarthy. My CV is available in postscript or PDF formats.
As researchers in the field of logical artificial intelligence, our main objective is to formalize common sense reasoning using mathematical logic. What does this mean? Common sense reasoning is the sometimes-unsound, but rather efficient, reasoning people use every day in order to accomplish tasks. It is not quite the same as using deduction, but we believe that we can formalize this different type of reasoning within logic.
Why is this important? We want to have robots perform all sorts of complicated tasks for us. In order for these robots to perform even the most basic tasks, they will have to reason. We first argue that they should be able to accept information and work out their goals themselves, as opposed to being directly programmed. This is the idea behind McCarthy's Advice Taker in [McCarthy 1959].
Our second argument is that the information the robot uses be in the form of symbols, or a logic. The alternate is non-declarative or non-symbolic representations such as neural nets, Bayes Nets, etc. We believe first-order logic (perhaps with some sprinklings of second-order logic) is a powerful and extensible representation language, that has the essential properties of elaboration tolerance, and expressivity.
I have also taught many classes, including:
I have also helped coordinate the Virtual Worldwide Seminar on Context for (1998--1999). This was a seminar that took place electronically at Stanford, Italy, Scotland, Sweden, and other areas of the world. Massimo Benerecetti and I developed the protocols so that all one needed to participate was a speakerphone. We would like to start up a Logical AI one at some point, given enough interest.
For 1999--2000 I co-coordinated the Broad Area Colloquium on AI, Geometry, Graphics, Robotics, and Vision. This is a weekly, prestigious colloquium to which established researchers are invited to present their thoughts.
I have put up a FAQ on admission to the Stanford CS department.
My academic history so far is short and sweet. I graduated from MIT with a Bachelor of Science in Computer Science in 1996. I continued my research on sound-to-letter/letter-to-sound generation systems with the SLS (Spoken Language Systems) research group at MIT. I received a Master of Engineering in Electrical Engineering and Computer Science based on this research in 1997.
A thesis was required for the Masters Degree. This thesis describes ANGIE, a system which can parse both spellings and phonetics of words into a hierarchical, probabilistic framework. ANGIE is used to automatically generate phonetic lexicons used in speech recognition. The thesis addresses the addition of sub-word elements, called morphs, to the framework, to improve transcription accuracies. It also introduces a tool that I developed to help automate the process of transcription.
During my time at Stanford I have been very fortunate, not only to work for John McCarthy, but also interact with some of the smartest people I know, including Eyal Amir, Dan Boneh, Tom Costello, Solomon Feferman, Richard Fikes, Mike Genesereth, RV Guha, Daphne Koller, Sheila McIlraith, Grigori Mints, and Johan van Benthem. The main focus was on aspects of common sense reasoning, knowledge representation, and logic. Some topics I have explored include planning, reasoning about action, stable model semantics, approximation, some natural language understanding, Hilbert epsilon semantics, the guarded fragment of logic, and greedoids. My dissertation is on elaboration tolerance, which I successfully defended on May 23, 2003.
I now work at Google.