Network Security Internet Technology Development Database Servers Mobile Phone Android Software Apple Software Computer Software News IT Information

In addition to Weibo, there is also WeChat

Please pay attention

WeChat public account


AI model can be used to solve complex discrete mathematical problems, Google DeepMind announced "FunSearch" training method

2024-04-23 Update From: SLTechnology News&Howtos shulou NAV: SLTechnology News&Howtos > IT Information >


Shulou( Report--, December 15 (Xinhua) Google DeepMind recently announced a model training method called "FunSearch", which claims to be able to calculate a series of "complex problems involving mathematics and computer science", including "upper-level problems" and "packing problems".

▲ source Google DeepMind (the same below) it is reported that the FunSearch model training method mainly introduces a "Evaluator" system for the AI model, the AI model outputs a series of "creative problem-solving methods", and the "evaluator" is responsible for evaluating the problem-solving methods of the model output. After repeated iterations, an AI model with stronger mathematical ability can be trained.

Google DeepMind uses the PaLM 2 model for testing, and the researchers set up a dedicated "code pool", enter a series of questions for the model in code form, and set up the evaluator process, after which the model will automatically select questions from the code pool in each iteration, generate a "creative new solution" and submit it to the evaluator for evaluation, in which the "best solution" will be re-added to the code pool. Restart another iteration. noted that the FunSearch training method is particularly good at "discrete mathematics (Combinatorics)". The model trained by the training method can easily solve extreme combinatorial mathematics problems, and the researchers introduced the process method of model calculation "upper bound problem (a central problem in the field of counting and permutation in mathematics)" in the press release.

In addition, the researchers also successfully used the FunSearch training method to solve the "packing problem" (Bin Packing Problem), which is a problem of "putting items of different sizes into a minimum number of containers". FunSearch provides a "real-time" solution to the "packing problem" and generates a program that "automatically adjusts according to the existing volume of the item".

The researchers mentioned that compared with other AI training methods that use neural networks for learning, the output code of the model trained by FunSearch training method is easier to check and deploy, which means it is easier to integrate into the actual industrial environment.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.

Views: 0

*The comments in the above article only represent the author's personal views and do not represent the views and positions of this website. If you have more insights, please feel free to contribute and share.

Share To

IT Information


© 2024 SLNews company. All rights reserved.