Make a fortune with your little hand, give it a thumbs up!
Today, Google DeepMinduses deep reinforcement learning to discover faster sorting algorithms, and the results of related papers [1] have been published on Naturethe Internet. According to reports: the algorithm can speed up 70%, compared to 3 times faster.
Summary
Basic algorithms like sorting or hashing are used trillions of times on any given day. As computational demands grow, it becomes critical to make these algorithms as efficient as possible. Given the remarkable progress made in the past 2 years , further improving the efficiency of these routines has proven challenging for both human scientists and computational methods. Here we show how AI can surpass the current state of the art by discovering hitherto unknown routines. To achieve this, we made the quest of finding a better sequencer a single player game. We then trained a new deep "reinforcement learning" agent, AlphaDev, to play this game. AlphaDev discovers from scratch small sorting algorithms that outperform previous known human benchmarks.
How can deep reinforcement learning improve the performance of ranking algorithms?
Deep reinforcement learning trains a new deep reinforcement learning agent to play the game by transforming the performance-boosting task of a ranking algorithm into a single-player game. The agent discovers small sorting algorithms from scratch that outperform previously known human benchmarks. In this way, deep reinforcement learning can go beyond the current state of the art and discover hitherto unknown ranking algorithms.
Application direction
These faster sorting algorithms can be applied in various real-world scenarios, such as search engines, databases, financial transactions, and other fields that require a large number of sorting operations. In these domains, faster sorting algorithms can improve computational efficiency and responsiveness, thereby improving overall system performance and user experience. In addition, these faster sorting algorithms can also be applied to tasks that require large-scale data processing and analysis in the fields of artificial intelligence and machine learning.
! career thinking
Will new algorithm professions emerge in the future, through deep learning, update and iterate existing algorithms, or use artificial intelligence to discover and create new algorithms?
With the continuous development of deep learning and artificial intelligence, it is foreseeable that new algorithmic careers will emerge in the future. Here are some possible directions:
-
Algorithm Engineer/Researcher: This is a broad field encompassing professionals who develop, improve, and optimize various algorithms. They will use deep learning and other related technologies to update and iterate existing algorithms and design innovative algorithms to solve new problems. -
Generative AI experts: Generative AI experts will research and develop generative models and create new algorithms through deep learning. They will explore how generative models can be used to generate new ideas and solutions in fields such as natural language processing, image generation, audio synthesis, and more. -
Reinforcement Learning Expert: Reinforcement learning is a machine learning method that learns the best behavioral strategies by interacting with the environment. Reinforcement learning experts will work on developing and improving reinforcement learning algorithms so that they can be applied in various fields such as autonomous driving, robot control, game strategy, etc. -
Automated algorithm optimization experts: Professionals in this field will use deep learning and other optimization techniques to develop automated algorithm optimization tools. Their goal is to improve the efficiency and performance of algorithms, leading to faster and more accurate problem solving. -
Data Scientist: Data scientists will use deep learning and artificial intelligence techniques to discover and create new algorithms. By analyzing and modeling large amounts of data, they will look for patterns, extract features, and use this information to build new algorithmic solutions.
These are some of the algorithmic career directions that may emerge in the future, but as technology continues to advance and applications continue to expand, other new careers and opportunities will emerge. The key is continuous learning and staying abreast of technological developments in this rapidly changing field.
Reference
Nature: https://www.nature.com/articles/s41586-023-06004-9
This article is published by mdnice multi-platform