Main Page

From Matanya Horowitz Wiki

Jump to: navigation, search
Matanya
Matanya

What creates intelligence? If we can't create it, how do we simulate it? What is the best way to perform a task? What will it take to actually bring robots not only into more factories but into service and the home? What are the tools we can use to get us there? How will these things be created, and how will they think?

These questions, and many others, are core to fields of economics, electrical engineering, controls, robotics, computer science, and mathematics. Each of these subjects has a different approach, and at this larger intersection there is a great deal left to do, but we're always getting closer to a new kind of intelligence.

I am a graduate student at the California Institute of Technology in the Department of Controls and Dynamical Systems. I recently graduated from the University of Colorado, Boulder with BS degrees in EE, CS, Appl. Math, Econ, and an MS in EE.

Contents

What is math research?

Interestingly, nearly nobody really understands what researchers in these fields actually do, even though they are creating some of the most exciting new technology. One project of mine is what might be called a controls-oriented approach to analyzing flow stability in complex networks. What this means is that I'm helping figure out how to speed up the internet. Did you know the internet doesn't work very well? Other subjects, like exact matrix completion, really means figuring out how to make videos really small. There are plenty of other examples. What I want to do is make the internet faster, make robots think quicker, and make a program that can play Starcraft - a computer game.

Do those sound like minor points? They're actually very interesting. Let's look at Starcraft. If you're successful you will have a machine that can successfully plan to build an army, tell its soldiers what to do, and can predict decisions. This isn't so different from a robot walking around your house that plans its day, knows how its decisions are going to affect your mood, and how to save time. The last piece would be how to fold the clothes and how to recognize the stuff around your house - problems involving what is called simulation and machine vision. You start putting these problems together, each individually sort of interesting (like Starcraft) and soon you've built something that seems intelligent.

Total Control

A more intelligent, perfect, scalable control of large scale systems, such as the Internet
A more intelligent, perfect, scalable control of large scale systems, such as the Internet

There has been a push in recent years, through such topics as game theory and motivated by tractability concerns, towards distributed control of large systems. However, at the same time, because wireless communication and other communication/sensing technologies have only become cheaper, there is also now the capability for coordination on an unprecedented scale. What if we could coordinate all the traffic lights in an entire city, and since we are looking at it all at the same time, do it perfectly? How are we going to manage our swarms of robotic insects? Large scale trading systems, army size logistics, internet traffic coordination, and a progressive automation of entire industries?

The difficulty is that these problems are large, and if you were to give them to existing algorithms they would take an unreasonable time to solve (order of millions of years or so, depending on the problem). The trick will lie in being able to break up these problems and compute them in parallel.

These are the types of problems I work on.

Some interesting stuff

Contact

Mailing

  • Matanya Horowitz
    California Institute of Technology
    1200 East California Blvd
    MC 305-16
    Pasadena, CA 91125

Electronic

Some recent stuff

Retrieved from "http://matanyahorowitz.com/index.php?title=Main_Page"