Democratizing IT platforms and tools is making these platforms and tools available to all users, regardless of their skill level. Often, a simplified interface is used or the complexity behind it is removed and a new layer of visualization is added.
Productising is a similar concept in some ways. However, it involves bundling several sub-technology elements that were not previously integrated, and creating new, larger and more important products or services beyond simple integration.
And there is an Operationalizing or Operationalization process, which can be said to be a combination of democratization and commercialization. It’s about making IT tools available faster where they need it. They often implement and provision custom quality levels in advance so that they can be consumed immediately. It’s like a sandwich.
Whether you’re skeptical of new IT concepts or open-minded about becoming readily available, you need to become familiar with the IT expressed in these concepts and terms. That way, in terms of software engineering, you can differentiate between completely new and just new packaging of old products.
Division through’banana logic’For example, consider the first development in 2021 of the Splice Machine in San Francisco, USA. Splice Machine is not a company specializing in banana and ice cream desserts. It is a data science specialist integrating AI/ML into modern cloud applications.
Splice Machine recently enhanced its platform to simplify the’feature engineering’ process behind adding AI/ML capabilities to applications. ‘Functional Engineering’ is the process of extracting features from raw data through data mining techniques using domain knowledge.
Nowadays everything is done in its own store or marketplace, and the splice machine is no different. This new service is called the’Splice Machine Feature Store’. This kind of’store’ reduces the complexity of functional engineering, and serves as a’shortcut’ to operating machine learning so that data scientists can make the right decisions based on real-time data.
What is the reason for doing this? According to a research report published by MIT in 2019, 70% of IT departments that have invested in AI/ML have no or negligible impact on how applications improve. Creating, developing, and elaborating machine learning models is a separate thing from making them operational in an enterprise environment.
“The ability and ability to create, share, explain, and reliably reproduce functions for specific models is critical to the success of the data science team,” said Monte Zuben, CEO of Splice Machine. In the case of the existing method, data science operations were not scalable.” The Splice Machine’s feature store enables companies to use complex analytics in real time and turn real-time data into functions. In other words, there is no case where the model cannot receive information. It also stores a history of features that create a series of learning courses with one click.
Functional engineering is the most time-consuming and costly task in the data science cycle. Juven explained that while companies are trying to make machine learning operational, data science is too productive to make it widely applicable.
“Simplifying data science workflows by providing the necessary architecture and automating functions with a function store are the two most important ways to make machine learning easy, accurate and fast.”
AI/ML operation becoming a trendIn addition to Spice Machine, there are many IT solution companies trying to increase the accessibility and convenience of AI/ML. Companies like Infor are using the Coleman AI toolset to market AI features on industry-specific ERP platforms.
In the case of IBM, for the past five years, we have been talking about machine learning automation and operation via Watson AI. In 2018, IBM unveiled NeuNetS (Neural Network Synthesis), which basically provides a way to synthesize neural networks by designing custom designs for specific data sets. In other words, it provides a more’smarter’ shortcut.
Salesforce doesn’t use the term operationalization a lot when it comes to Einstein AI. Instead, it focuses on the broader term’customer success platform’. But in many cases, it is the same concept. Accenture uses the term Applied Intelligence, which is also in the same context.
How to operate AI/MLPerhaps the real dynamics of AI/ML operations, which themselves require discussion and analysis, entail some consideration of significant key technologies. The software’smartization’ that has been actually operated includes function reuse and redundant function engineering removal. It also includes tools to help address governance issues, such as bias and regulatory oversight.
In order to run operational machine learning, the intelligence engine you use must have good training data. Scaling with machine learning tools requires insightful predictions about scalability provisioning. This is because sufficient cloud processing power and storage are required.
Democratization, operationalization, and commercialization will be part of IT over the next 10 years. Operationalization has already reached the stage of’normal’. [email protected]
Source: ITWorld Korea by www.itworld.co.kr.
*The article has been translated based on the content of ITWorld Korea by www.itworld.co.kr. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!
*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.
*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!