Yandex, the Russian search engine giant, has announced an open source machine learning library for gradient boosting called CatBoost. The new development is designed to teach systems and decrease the transactional or historical data.
The Google counterpart in Russia had already been evolved into various business lines similar to search giant. But the latest move has extended its reach from just a search engine to a machine learning support. “By making CatBoost available as an open source library, we hope to enable data scientists to achieve top results with least efforts, catalyse future innovation and ultimately define a new standard of excellence in machine learning,” said Misha Bilenko, head of machine intelligence and research, Yandex.
Yandex has currently been using MatrixNet machine learning algorithm across all its products and services. This innovation will be replaced with CatBoost in the next couple of months. The second big plan by Yandex is to offer CatBoost as a free service under Apache license. Any tech product wanting to use gradient-boosting tech in their own programs can consider using MatrixNet.
CatBoost is touted to provide “highly accurate results” even with relatively little data. It can work on small data sets of domains like sensory, transactional and historical data in addition to supporting a data formats including inputs provided by deep learning models and sensory data provided through images, audio and text.
No plans to commercialise
While Yandex is expecting to grow its presence in the artificial intelligence space with the launch of CatBoost, it has no plans to commercialise the algorithm in any proprietary way. However, the Moscow-headquartered company is likely to replace MatrixNet with CatBoost in Yandex.taxi service in coming few months. That service was debuted as a joint venture with Uber across Russian market.
Developers can access the CatBoost code through its official GitHub repository. The library is based on JavaScript as well as C and Jupyter Notebook languages.
