Looking for a good problem for a solution I have coded :)

  • Thread starter Thread starter tomekd789
  • Start date Start date
Click For Summary

Discussion Overview

The discussion revolves around finding suitable problems for a genetic algorithm-based neural network abstraction developed by a participant. The scope includes practical applications of the tool, exploring various problem domains such as number factorization, satisfiability, and prediction tasks.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested
  • Mathematical reasoning

Main Points Raised

  • One participant has developed a genetic algorithm for breeding neural networks and is seeking non-trivial, practical problems to solve with it.
  • Previous attempts included number factorization, checking satisfiability of random 3-SAT terms, and predicting Forex trends, with varying degrees of satisfaction.
  • Another participant suggests that searching for a problem without a clear objective may not be cost-effective.
  • A participant shares an experiment predicting the results of manually generated coin tosses, achieving a 70% accuracy rate, and notes the simplicity of the network structure required.
  • There is a query about the effectiveness of the neural network compared to a simple strategy of predicting the opposite of the previous coin toss.
  • A participant questions the significance of the results and suggests exploring whether the algorithm can solve linear systems or learn from data.
  • Suggestions for potential problems include exploring genetic algebras and decryption tasks, starting with simpler codes like the Caesar cipher.

Areas of Agreement / Disagreement

Participants express differing views on the effectiveness and significance of the neural network's performance, and there is no consensus on the best problem to pursue.

Contextual Notes

Some participants express uncertainty about the significance of the results and the capabilities of the algorithm, particularly regarding its ability to solve linear systems or learn effectively.

Who May Find This Useful

Individuals interested in genetic algorithms, neural networks, and practical problem-solving in computational contexts may find this discussion relevant.

tomekd789
Messages
3
Reaction score
1
I have coded the following abstraction for breeding neural networks with a genetic algorithm: https://github.com/tomekd789/clogann
Now I am looking for a problem to solve using it. It would be ideal if it is non trivial, of practical significance, and matching the tool. So far I've tried:
  • Numbers factorization (with a mediocre effect)
  • Checking satisfiability of random 3-SAT terms taken from the phase transition area (http://www.princeton...s/gent94sat.pdf) (somehow dissatisfied; did not reach the goal of proving P = NP http://cdn01.codecall.net/public/style_emoticons/default/wink.png )
  • Forex (with positive effects, but not worth the effort).
All hints welcome.

Regards,
Tomasz
 
Last edited by a moderator:
Technology news on Phys.org
Solution looking for a problem is often not cost-effective.
 
Anything more concrete? ;)
Meanwhile, I've been advised to take a sequence of "coin tosses" generated manually by a human (i.e. no real coins, no computer involved, just from one's head) and breed a network capable of predicting next tosses results. I was astonished: for 1000 "coin tosses" the network was 70% correct.
I.e. we humans are quite predictable, at least in this case. :)
Another surprise was the necessary minimal size of the network: just two input neurons, and three extra, one treated as output, too.

Still looking for other ideas.
Tomasz
 
Which ratio do you get if you simply predict "the opposite of the previous coin"? Is the neural net better than that?
 
65.2%, then still slightly worse.
 
  • Like
Likes   Reactions: mfb
At first glimpse this isn't very significant. What does your algorithm? Can it solve linear systems or can it learn?
Without being an expert I'd try to google 'genetic algebras' and see what kind of problems there are.
Or try some decryption problems. You can start with an easy Cesar code. Just to see how it performs. And then more difficult ones.
(I'd have a bunch of tasks but they all involve solving linear systems with free variables. They are ##O(n^3)## which quickly makes it boring to solve them by hand.)
 

Similar threads

Python How Can I Fix the IndexError in My 2D Ising Model Simulation?
  • · Replies 6 ·
Replies
6
Views
6K
Current and Future AI Threats to Humanity and the Human Response
  • · Replies 10 ·
Replies
10
Views
5K
Key papers in computational neuroscience
  • · Replies 1 ·
Replies
1
Views
12K
Two of my nonsense short stories
  • · Replies 3 ·
Replies
3
Views
3K