Samsung tops AI challenge using Reinforcement Learning

Reinforcement Learning gives an AI feedback for each decision it makes, theoretically speeding up training

Samsung has placed top in two global artificial intelligence machine reading Comprehension competitions, through its R&D hub Samsung Research.

The company's research arm came top at both MS MARCO, the Microsoft Machine Reading Comprehension Competition, and TriviaQA, hosted by the University of Washington.

Both tests are designed to evaluate AI algorithms, their ability to process natural language in human Q&As and for analysing text in different types of documents.

Machine reading comprehension tasks an AI algorithm with analysing data and finding an optimum answer to a query. Both tests that Samsung chose use natural language processing, requiring the algorithm to understand questions submitted by humans (MS MARCO, for example, uses questions submitted to Bing), as well as providing various types of written text.

Samsung's algorithm is called ConZNet, which uses the Reinforcement Learning (RL) technique, which is being utilised in several AI implementations today. Self-driving startup Wayve is using RL to teach an AI how to drive a car using only a single camera, for example.

RL is a ‘carrot-and-stick' approach to machine learning, where the algorithm receives feedback for each of its decisions; Google is using the same system for its Deepmind AlphaGo.

Jihie Kim, Head of the Language Understanding Lab at Samsung Research, said: "We are developing an AI algorithm to provide answers to user queries in a simpler and more convenient manner, for real life purposes. Active discussion is underway in Samsung to adopt the ConZNet AI algorithm for products, services, customer response and technological development."

Samsung says that MS MARCO and TriviaQA are ‘among the actively researched and used machine reading comprehension competitions' by universities and AI firms around the world. They are, however, open competitions that participants can enter at any time, so Samsung could be displaced in the near future.