Facebook AI Research Director Yann LeCun believes that Deep Learning will soon need a new programming language that is more flexible and easier to work with than Python. He didn’t say it was a necessity but this possibility according to him runs against very entrenched from researchers and engineers.
He was speaking to VentureBeat regarding this and he also said that aside from Facebook, development teams from Google are also working on similar projects to design an efficient language that can harness Deep Learning capabilities. He said, however, it wasn’t clear whether the developer community would be willing to leave Python and opt for a new framework or not.
According to the latest report by GitHub, Python is one of the most preferred languages for developers working on Machine Learning projects. It is also essentially the core of Facebook’s open source machine learning framework PyTorch as well as Google’s own TensorFlow.
While presenting his research paper at a conference, he said that there could come a time when supervised learning could allow machines to learn a vast amount of background knowledge, there may be a chance that machine could have its own common sense. To achieve such progress, however, he said we would need high-performance hardware which still isn’t available.
Do you think we need a new language for Deep Learning?
