Currently engaged in research with leading faculty in machine learning and unconventional computation on my campus; in this project we have developed a boolean satisfiability (k-SAT) solver that uses an Asymmetric Continuous-Time Neural Network to model an analog circuit approach for computing large combinations of possibilities to solving logical problems.

So far, we have been able to solve large variable problems (up to 10,000) in digital software using the algorithms developed, and will soon implement this on silicon chip technology. This could lead to many other problems being able to be solved in a relatively short, close to non-polynomial time, solving the classical P =/!= NP problem in theoretical computer science and computational complexity theory.

Another amazing application of my research is that if this problem is solved with k-SAT, every other NP-Complete problem can be reduced down to this and solved with k-SAT. This means, that tasks such as factoring extremely large prime numbers, the key behind public-key cryptography and RSA keys -- the backbone of network security in most of the world -- would be easily solved by the hardware I am developing.

We are publishing in 2015.

Built With

Share this project:

Updates