源之原味

这里, 到处都是: 用自学 AI 改造物流

 

这篇文章来自 nvidia.com。原始 url 是: https://blogs.nvidia.com/blog/2018/03/28/self-learning-ai-for-logistics/

以下内容由机器翻译生成。如果您觉得可读性不好, 请阅读原文或 点击这里.

One of the longest-running challenges in the logistics industry is finding the shortest routes.

First articulated in the 1930s, the “traveling salesman problem” seeks to deduce the shortest route connecting a group of cities to ensure optimal use of time and resources.

Karim Beguir, co-founder and CEO of London-based AI startup InstaDeep, told GPU 技术会议 attendees this week that GPU-powered deep learning and reinforcement learning may have the answer.

Previous efforts to address the traveling salesman problem include optimization solvers, heuristics and Monte Carlo Tree Search algorithms. But, according to Beguir, these approaches all suffer from the same shortcoming: They don’t learn.

“There is no experience which is being gathered from the problems that were solved before,” said Beguir. “With the computational investments you’re making to solve this problem, it would be nice if there were some learning.”

But Beguir said recent work in combining the speed of deep learning neural networks with the deliberate decision making of Monte Carlo Tree Search is leading to breakthroughs that will transform logistics.

He pointed to the AlphaZero program created by DeepMind, which combined neural networks, Monte Carlo Tree Search and the ability to learn from self-play to create what he called “a new type of AI champion.”

AlphaZero was developed to beat the best Go, Shogi and chess players in the world. But its implications clearly reach much further.

“This is happening with zero data,” said Beguir. “The only data is what the system generates by playing against itself.”

AlphaZero bridges the two cognitive processes Daniel Kahneman described in his book, “Thinking, Fast and Slow.” In the book, one process is the fast and intuitive reactions that dictate most of our lives. The other is the deliberate thinking used to solve more complex problems.

Inspired by this advance, InstaDeep is working on injecting Monte Carlo Tree Search into deep learning to apply AlphaZero-like capabilities to solving business problems. Running its models on an NVIDIA DGX-1 AI supercomputer is leading to some powerful developments.

Rather than having to be trained, InstaDeep’s AI algorithm works from scratch, solving the traveling salesman problem by finding progressively better paths. The algorithm also is learning how to pack bins more efficiently, addressing another logistics industry challenge.

While Beguir believes that creating AI that learns from experience is a better approach to solving critical problems in industries like logistics, he concedes there’s a ways to go before self-learning AI evolves into a mature business solution.

“This is just scratching the problem, and we expect to see a lot more on that front,” said Beguir. “A few years from now, if your system doesn’t have learnability in it, you’re probably doing something wrong.”

InstaDeep is a member of NVIDIA’s 初始程序, which helps accelerate startups pushing the frontiers of AI and data science.

Leave A Reply

Your email address will not be published.