IBM gives access to the main component of Watson
On February 15, IBM opened access to a core machine learning component used by the Watson supercomputer, which will allow other organizations to adapt the system’s capabilities to their needs. We are talking about the IBM Machine Learning platform, whose task is to reduce the complexity of developing and deploying specialized analytical models.
/ Flickr / Mr Seb / CC
Over the past few years, we have seen how IBM used Watson's intellectual capabilities in a wide variety of fields: from speech recognition to medical research . The technology has even been used to combat cybercrime. Such a platform was launched in test mode in December last year in order to identify behavioral patterns and other signs of cyber attacks. The solution was able to handle up to 80% of all information on the Internet, including blogs, articles, videos, etc.
')
Watson also showed itself while playing Jeopardy! To do this, the developers had to solve many problems of natural language processing, related to the classification and assessment of issues, building dependencies, knowledge representation, etc. For more information, please click here .
And last week there was an opportunity to apply the ability of Watson in other areas. “Now you can use the power of machine learning when working with any type of data: databases, unstructured data, emails, and so on,” said Rob Thomas, general manager of IBM Analytics.
One of the main features of IBM Machine Learning is a built-in recommendation engine that should help data analysis and processing specialists select algorithms for their projects. The mechanism assesses a number of parameters, including what kind of records the company wants to process and how quickly results are required.
Then the system looks for analytical algorithms in its database that are best suited for the task. After that, it gives scientists the opportunity to customize the formulas if the ideal solution could not be found, since the tasks for which the platform's power will be used are hardly trivial.
The developers say that IBM Machine Learning works with any programming language and supports most of the industry-popular AI frameworks, including SparkML, which IBM itself is actively developing. The company also proposes to use the SystemML tool , whose task is to optimize the performance of algorithms.
IBM Machine Learning will first be available on Systems Z mainframes, which are used by companies in regulated industries, such as financial institutions. Big Blue also hopes to add support for machine learning tools to make the platform as accessible as possible. Big Blue says IBM Machine Learning will enable organizations to more quickly build and deploy high-quality, self-learning models.
PS We suggest reading several materials on the topic from our blog:
PPS What else do we write about:
/ Flickr / Mr Seb / CCOver the past few years, we have seen how IBM used Watson's intellectual capabilities in a wide variety of fields: from speech recognition to medical research . The technology has even been used to combat cybercrime. Such a platform was launched in test mode in December last year in order to identify behavioral patterns and other signs of cyber attacks. The solution was able to handle up to 80% of all information on the Internet, including blogs, articles, videos, etc.
')
Watson also showed itself while playing Jeopardy! To do this, the developers had to solve many problems of natural language processing, related to the classification and assessment of issues, building dependencies, knowledge representation, etc. For more information, please click here .
And last week there was an opportunity to apply the ability of Watson in other areas. “Now you can use the power of machine learning when working with any type of data: databases, unstructured data, emails, and so on,” said Rob Thomas, general manager of IBM Analytics.
One of the main features of IBM Machine Learning is a built-in recommendation engine that should help data analysis and processing specialists select algorithms for their projects. The mechanism assesses a number of parameters, including what kind of records the company wants to process and how quickly results are required.
Then the system looks for analytical algorithms in its database that are best suited for the task. After that, it gives scientists the opportunity to customize the formulas if the ideal solution could not be found, since the tasks for which the platform's power will be used are hardly trivial.
The developers say that IBM Machine Learning works with any programming language and supports most of the industry-popular AI frameworks, including SparkML, which IBM itself is actively developing. The company also proposes to use the SystemML tool , whose task is to optimize the performance of algorithms.
IBM Machine Learning will first be available on Systems Z mainframes, which are used by companies in regulated industries, such as financial institutions. Big Blue also hopes to add support for machine learning tools to make the platform as accessible as possible. Big Blue says IBM Machine Learning will enable organizations to more quickly build and deploy high-quality, self-learning models.
PS We suggest reading several materials on the topic from our blog:
- "Neuromorphic chips": a different look at machine learning
- How neural networks are now used: from research projects to entertainment services
- "Everywhere at Home": What technologies does Airbnb use?
- Neurostream - a new chip for deep learning
- Friday format: What happens when Moore's law stops acting
PPS What else do we write about:
Source: https://habr.com/ru/post/322196/
All Articles