REI - Regular Expression Inference Forum

Go back to competition Back to thread list Post in this thread

> How to solve this problem mathematically ?

.

Posted by: vetrivel @ Jan. 24, 2024, 5:59 a.m.

Hi, and thanks for your interest in our challenge.

I'm not sure I'm able to answer this question; however, as we say in the paper, this is a very hard problem that requires a solver, which essentially enumerates (i.e., exhaustively searches) expressions to find the minimal RE for a given PN set and cost function. As far as we are aware, the only viable mathematical way is enumeration.
Refer also to "Minimizing NFA's and Regular Expressions" by Gramlich and Schnitger (https://www.sciencedirect.com/science/article/pii/S0022000006001735), and to "Search-Based Regular Expression Inference on a GPU
" by Valizadeh and Berger (https://dl.acm.org/doi/abs/10.1145/3591274), the latter of which is the solver we actually use to generate the data for this challenge.

Posted by: REIC @ Jan. 24, 2024, 3:51 p.m.

Hi,

It cant be possible that no one had solved this problem yet.

everyone are stuck with academics, or not seeking collaboration from mathematicians.

Computing is a practical machine created from theoretical maths. Any computer problem is math solvable.

Posted by: vetrivel @ Jan. 24, 2024, 4:21 p.m.

Hello,

If you have sources for known approaches that solve the REI problem mathematically (and efficiently), we would be very grateful if you could share them with us; this challenge aims to push the field of ML in this area, and any proven mathematical approach could help us and challenge participants :)
There are proven boundaries for this problem, that impose limits on any mathematical and computational approach; please see the papers we mentioned above for an overview.

Posted by: REIC @ Jan. 24, 2024, 4:48 p.m.

Hi.

ML and any learning programs are flawed fundamentally. Here is why.

Inputs + function = output. This is typical computing

Inputs + outputs = function. This is ML or other learning model.

Please also note, I'm trying to communicate few things.

1. Reading a lot will rot the brain.
2. AI ML is a trend not truth.
3. This problem, where does it have practical impact? How can I collaborate with problem setters in person,? A whiteboard and face to face interaction, I can solve it mathematically, and a working solution computationally.

Please also note, solve first, making efficient comes next because once it I solved, kick is gone.

Posted by: vetrivel @ Jan. 24, 2024, 5:16 p.m.

Just curious, what is the practical use of solving this problem ?

Posted by: vetrivel @ Feb. 1, 2024, 12:02 p.m.
Post in this thread