Yandex, the Russian Google competitor, has for a while been quietly offering its MatrixNet machine learning technology to other organizations – CERN, for example, has used it to establish statistical relevance in its floods of physics data. On Tuesday, Yandex announced a more formal push into offering big data services to corporate and enterprise clients.
The company revealed Yandex Data Factory at the Le Web conference in Paris. Yandex said its technologies can be used for the personalization of recommendations, natural language processing, image and speech recognition, credit scoring, logistics optimization, demographic profiling and so on.
According to Yandex, these services have already been used by a leading European bank to crunch behavioral data, so as to match products to specific marketing channels in a personalized way. The firm’s machine learning and geolocation services were also used by a road management agency to boost accident prediction accuracy. All in all, Yandex is already providing big data services for 20 projects.
Yandex says it’s able to create various kinds of deep neural networks with MatrixNet, which is used to train ranking formulas, boosting the effectiveness of the learning process. The firm claims the cluster management tools in its Friendly Machine Learning framework make it easier to get into big data research by allowing researchers to “avoid dealing with distributed computing problems.” It uses proprietary technologies including Yandex MapReduce and Real Time MapReduce, and Yandex Tables, which is a big data storage and processing platform.
Yandex Data Factory has an Amsterdam office as well as one in Moscow, and also has data centers across various countries. The company also offers training programs for corporate tech teams and free masters-level courses for university graduates — it’s actually had a data analysis school since 2007.
The division’s chief, Jane Zavalishina, said in a statement:
Yandex Data Factory uses algorithms that Yandex developed for its own needs: search, traffic forecasting, ad targeting, music recommendations. But it’s not the ‘content’ of data that these algorithms analyse – rather, they analyse data interrelations – and so they can be applied in any industry: from banking to telecommunications, from logistics to oil and gas extraction, from public utility services to aircraft engineering.