Real-time AI systems are platforms capable of processing requests as fast as they receive them. Implicitly, this means that the respective AI systems should be able to learn in real-time.
However, we’re still a long way from reaching this ambitious goal. Remember that line from Star Trek: “Space – the final frontier”? Well, in AI research that line goes like this: “Real-time AI learning: the final frontier”.
How AI learns
Machine learning is the cornerstone of Artificial Intelligence. But the machine learning process is a static one and it requires a lot of time. First, there is the training phase where researchers use huge corpora and piles of data and only then comes the main process of testing and adjusting that model in practice. AI is highly dependent on the initial training set.
Meet Project Brainwave
Project Brainwave is Microsoft’s newest deep learning acceleration platform. This powerful hardware for machine learning can process 39.5 teraflops in machine learning tasks in less than 1 millisecond without batching tasks together. This means that Project Brainwave is capable of handling extremely complex AI tasks on the spot.
These new AI chips are more flexible than regular CPUs and, most importantly, can be reprogrammed for different tasks. As a quick reminder, standard CPUs are very rigid and need to be replaced with new hardware if users want them to perform a different task.
The Redmond giant is planning to integrate these new AI-powered CPUs in its Azure cloud platform in the future.
Real-time AI learning
But there’s more to Project Brainware than meets the eye. These new chips rely on deep neural networks. This AI technology allows computers to process information like humans do.
So, what does this mean? Project Brainware lays the foundation for real-time AI learning. These new processors could soon support machine learning systems that run in real-time, without any hardware constrains.
Thanks to their flexibility, researchers can use them in virtually all machine learning fields, including machine translation, speech recognition, and more.
Speaking of which, let us take an example. Let’s image that you use a word that your automated translation software can’t find in its dictionary. These CPUs could allow the software to infer the meaning of the unknown word using the information available in the context. Of course, this is only one of the many possibilities that these AI CPUs open.
YOU MAY ALSO LIKE: Google creates AI agents that can imagine and planFollow The AI Center on social media:
Join me as I track the latest progress in AI research.
Latest posts by Maddie Blau (see all)
- True Emoji is an AI app that uses your expressions to create animated emojis - November 21, 2017
- Canada’s first AI exchange-traded fund enters the market - November 2, 2017
- Are you curious to see who’s the smartest AI in the world? - November 1, 2017