Prof. Sussman's Reading List

aurellem

Written by:

Gerald Sussman (compiled by Robert McIntyre)

Serving as a TA for Professor Sussman will get you three things: great advice, spectacular reading recommendations, and lots of high quality tea. I can't share the advice or the tea, but I can compile a reading list. Some of the materials on this list represent research paths that lead to unexplored territory. Some are textbooks that express concepts so clearly they will change your life and make you weep for joy. I hope that you will get something interesting out of this reading list, wherever you are in life – there's stuff I wish I knew about in middle school, and there are things I can't wait to read this summer. Enjoy! (and send corrections to reading-list@aurellem.org!)

–Robert McIntyre

1 Some Real High school Reading

2 Representative Student Theses

These are students where I played a large role in their education. Many of them represent compelling research directions that desperately need to be extended by the next generation of researchers! As Minsky says, if you want to do something really new, go back to points in the past where there was a neat idea that never really caught on, and follow the path of that idea to see where it leads. A comprehensive list of all my student's works can be found at my homepage.

–Prof. Sussman

In particular, here's two great ideas that seem extremely promising and have NOT been properly explored! You could be the first person to get them working!

  • Using chaos to get unlimited measurement precision!
    • In chaotic systems, states that are near to each other at one point in time become exponentially farther apart from each other as the system evolves in time. Therefore, you might be able to attain arbitrary precision by waiting for the system to evolve, and then determining what initial state must have led to the later state.
    • Two notable papers:
    • No one put enough effort into seeing if it really worked.
    • Seems to allow for almost unlimited precision in measurement.
    • Initial results look very promising, with a 13 order of magnitude improvement in measurement precision in a simple experiment.
    • You will win the Nobel Prize if you can get it to work, because you will revolutionize the way we do measurements. In particular, you could measure the Gravitational Constant with unprecedented accuracy.
  • Towards Intelligent Structures: Active Control of Buckling
    • By Andrew A. Berlin, 1994
    • Achieves a 10 fold increase in strength by actively eliminating vibrational modes.
    • Such a good idea; It's cool, short – great!
    • Better quality, color version of the thesis here.
    • No one's followed up on it!

In historical order:

3 From Sussman's Bookshelf:

4 Marvin Minsky

Minsky really made me as a person. He was my adviser when I was a student at MIT, and he got me my first job. He had the "magnetism" to attract the most talented people to MIT to work on AI, and the right amount of negligence and delegation to create an environment where people could thrive. He is certainly the reason that I was seduced into working on AI. Minsky has vast and deep Scientific knowledge – he could walk into almost any class: Chemistry, Physics, Math, Computer Science, and teach the class well without preparation!

–Prof. Sussman

  • http://web.media.mit.edu/~minsky/ Much of Minsky's work is here, including his book, The Emotion Machine, and several essays and papers. Check it out!
  • Society of Mind Read it online! Each chapter of this book is a short, self-contained essay about some aspect of intelligence or development.
  • Music, Mind, and Meaning Minsky is one of a few living people who can improvise complicated Baroque era fugues. You can hear one of these improvisations here.
  • Steps towards Artificial Intelligence Here, Minsky outlines how we might begin to build an AI. This is considered to be one of the founding papers of the field, along with Turing's "Computing Machinery and Intelligence" paper.
  • Perceptrons, by Marvin Minsky
    • MIT Press, ISBN: 9780262631112
    • Really good for "Math types."
    • Uses geometry for proving things.
    • People unwisely considered it to kill off Neural Nets; In fact, it only shows the limitations of certain simple kinds of Neural Nets.

5 For Fun

6 Selected works by Sussman

Author: Gerald Sussman (compiled by Robert McIntyre)

Created: 2016-07-13 Wed 00:16

Emacs 24.5.1 (Org mode 8.3beta)

Validate