Baidu releases open source AI code
Fri 15 Jan 2016
Baidu, a massive Chinese web company along the lines of Google, has released artificial intelligence software WARP-CTC on GitHub. WARP-CTC, developed at Baidu’s Silicon Valley AI lab, was created to improve speech recognition in Baidu’s end-to-end speech recognition program Deep Speech 2.
WARP-CTC functionality in Deep Speech 2 has been shown to allow speech recognition in English and Mandarin, and in some cases the program is better at speech recognition than humans. Andrew Ng, Chief Scientist for Baidu, stated, “For short phrases, out of context, we seem to be surpassing human levels of recognition.”
WARP-CTC builds upon an AI fundamental called connectionist temporal classification. Finding existing CTC AI functions slow and cumbersome, the team at Baidu parallelized the CTC algorithm for increased speed and functionality. The WARP-CTC software released today includes a simple C interface and the bindings for Torch, a scientific computing framework. This should allow users to easily incorporate WARP-CTC into existing deep learning projects. Company Nervana is also incorporating WARP-CTC into their program neon, a Python-based deep learning library.,
Baidu seems to be following in the footsteps of Google, which released an open-source AI program in November called TensorFlow. TensorFlow is also used to improve speech recognition as well as a number of AI applicable functions – web searches, translation and more. The idea was that Google, as a large corporation heavily invested in research and development, was years ahead of the curve in terms of AI advancement. By sharing their code as an open source, and providing these tools to interested parties, they could advance machine learning world wide.
The same philosophy exists at Baidu. At a Q&A on their website, the company wrote: “we want to make end-to-end deep learning easier and faster so researchers can make more rapid progress … we want to start contributing to the machine learning community by sharing an important piece of code that we created.”
Baidu has encouraged developers using the open source WARP-CTC to reach out to the company, and expects to release more AI tools in the future.