IBM Watson Studio - a cloud platform for developing AI applications

Introducing IBM Watson Studio

Experts are beginning to use artificial intelligence to transform our most valuable resource - data - into new forms of business. Artificial intelligence gives us the opportunity to end the battle with the data and start using them to prepare convincing recommendations, accelerate research and improve the quality of interaction with customers in a convenient way. The task of artificial intelligence systems is to complement human intelligence, and today we are taking another step towards making AI more accessible to everyone. Meet IBM Watson Studio .

Watson Studio: useful for organizations working with artificial intelligence

Watson Studio speeds up the workflows associated with machine and deep learning needed to integrate AI into business and support innovation. The solution includes tools for data analysts, application developers, and specialized experts to help organize collaborative data processing and large-scale layout, training, and model deployment. For successful implementation of AI projects, algorithms, data, specialists and a very powerful computing infrastructure are needed.

Until today, there was a barrier between data experts and specialized experts. Only technicians of higher qualification could systematize and analyze huge amounts of data. Only profile experts could transform the data into the knowledge needed by artificial intelligence. However, they worked independently of each other, used different tools and could not share the results of work. As a result, artificial intelligence could not really complement the human intellect.

Watson Studio eliminates this barrier and allows you to create a unified environment for generating new ideas based on the knowledge contained in the data. Watson Studio provides an opportunity to organize the interaction of experts of different profiles throughout the organization. Experience working with clients around the world convinced us that complex interaction is the key to realizing the full potential of AI.

A complete set of tools for integrated application development using AI

Watson Studio contains a variety of tools for organizing the full life cycle of AI, including best-in-class open source tools, as well as tools from IBM. Functional components may or may not require the development of code and provide an opportunity to compose and train their own machine learning (MO) and deep learning (GD) models, as well as re-train and configure pre-trained Watson APIs. Extensive options are available for fine-tuning models, managing them and automating the feedback loop, which allows models to become smarter over time and constantly adapt to changing external conditions.

Connection and data preparation

The use of artificial intelligence begins with connecting and downloading data. Data analysis experts spend up to 80% of their time searching and preparing data; at the same time, 57% of experts indicated during the survey that cleaning and systematization of data is the most unpleasant part of their work. And not only they face this problem. Business analysts have similar difficulties with obtaining the data necessary for the preparation of reports. Sometimes they have to wait several weeks for IT staff to extract the necessary data from the source systems.

To solve the problem of cleaning and processing data, we have developed a Data Refinery tool that allows you to organize a quick independent data preparation. Watson Studio includes over 35 interfaces for connecting to data sources from the IBM Cloud, other vendor clouds, applications, and local systems.

Tools and trained Watson models

After connecting to the data you need to build and train the model. Application developers can start with proven, trained and optimized Watson APIs. These models understand the mood, classify topics in the text, identify personality traits, and recognize objects in photographs. We provide access to a fully documented API with examples and code snippets in common programming languages.

Not all companies have the expertise and resources to develop innovative machine learning and in-depth training technologies. Therefore, we offer simple tools to help fill Watson with knowledge from your own data. The first built-in tool is Watson Visual Recognition. It allows you to train special models using your own images to solve individual problems in the field of pattern recognition.

Framework selection and best-in-class tools

Machine and deep learning is constantly evolving. Watson Studio supports the most popular tools for teaching, saving, deploying and automating the re-learning of models. These tools are part of the product, and our solution provides infrastructure management so you can concentrate on your projects.

Use Python, R and Scala in Jupyter Notebooks . Laptops are a popular environment for developing and exchanging documents with code, mathematical calculations, visualization, and comments. They are used for data cleaning and transformation, numerical modeling, statistical modeling, machine learning and more.

Together with the RStudio team, we created the most popular open source R statistical computing environment in the world. Watson Studio includes the flagship product RStudio, a popular development environment that simplifies data analysis using the R language.

Visual modeling tools for those who do not like to program

Although the open source Python and R environments are popular due to their low cost, flexibility, and extensive features, quality programming and code testing can take a lot of time and take a lot of effort. We interviewed many clients and realized that not all of them are programmers. Therefore, we offer two tools for those who prefer visual modeling:

• SPSS Modeler : a simple and user-friendly interface targeted at everyone - from business users to data analysis experts. It helps to quickly extract valuable ideas by creating visual models with the mouse. Thousands of Modeler users can import SPSS streams into Watson Studio, and they will work correctly!
• Neural Network Modeler : a simple and convenient graphical interface for developing neural network structures. This interface speeds up development, eliminating the need to manually write and debug code. Neural networks can be exported to TensorFlow, Keras, PyTorch and Caffe, as well as in JSON format for posting on blogs and uploading code to GitHub.

Flexible computing environments for large-scale model learning

Current AI models are highly complex and may require terabytes of data for training. Active work is underway to increase the computing power of processors, graphics cards, distributed learning tools, and other means. Watson Studio gives you the opportunity to choose the hardware for experiments and train models with maximum speed. You can quickly scale computing resources and customize package dependencies in computing environments.

Each computing environment is allocated to a specific project participant so that it does not need to compete for resources. Our solution supports package management — definitions of popular environments are applied to customize conda. Computing environments are treated as project resources and can be used by all team members for replicable research .

Because the computing environments are in the IBM Cloud, you only pay for what you use. Our flexible infrastructure easily copes with any peak loads and is designed to support the most demanding projects.

Model lifecycle management

Deploying the model in a production environment is usually a daunting task, requiring the assistance of busy IT professionals. When a deployment can take several weeks, it is not surprising that data analysts prefer to leave their newest model and move on to the next project instead of re-training and deploying existing models.

In Watson Studio, you can automate model re-training and monitor model performance over time. This continuous learning is a unique advantage of our platform. You can set thresholds, and if performance drops, users will receive warnings and notifications so that the data analyst can take the necessary action.

Models are dynamic resources that need regular updating. Therefore, it is very important to have version control tools that allow returning to previous versions if necessary. These tools must be accessible through the API and user interface.

Fast sharing of results using dashboards

Watson Studio dashboards simplify the visualization of data in applications. Users can independently create visual models to search for new ideas.

Interactive dashboards provide real-time data visualization. Smart tools for data analysis and visualization help to find patterns and meaning in the presented data. You can also use filters and data navigation tools. For maximum user experience, the dashboard users can be integrated into the application context.

Integration with Watson Knowledge Catalog

Watson Studio solution is designed to support the most demanding tasks that can be implemented using artificial intelligence in the organization. In particular, the solution includes an intelligent directory service that allows you to integrate and prepare analytical resources, including structured and unstructured data, wherever they are (in the local infrastructure or in the cloud), to increase the productivity of data analysis, machine learning and artificial intelligence.

Combine all information resources into a single directory with extensive metadata . Watson's artificial intelligence will recommend optimal resources based on Watson’s understanding of the relationships between resources and what users do with them.

A powerful integrated mechanism for activating data policies provides automatic protection of confidential information in accordance with the rules established in your organization.

Even more power for AI

Watson Studio gives models a chance to fulfill their mission: to learn. Continuous training of models using the latest data ensures that models are constantly matching changing circumstances and provides the organization with the knowledge necessary to make more informed decisions and realize competitive advantages.

Watson Studio is a product of close collaboration between IBM Research and other IBM Cloud teams. We hope that you enjoy our product, and look forward to the opportunity to find out what you will create with it.

Try Watson Studio in the IBM cloud!

You can try Watson Studio in the IBM cloud for free - register in the IBM cloud and try IBM Watson Studio:
https://www.ibm.com/cloud/watson-studio .

Useful resources

Studio Watson IBM youtube the playlist
Subscribe to the newsletter notifications free IBM seminars.
Materials to do lab work using IBM Watson Studio.

IBM Startup and Partner Support Programs

In addition to the opportunity to try IBM services, our company offers programs to support companies that want to use the IBM cloud for testing and deploying applications. These programs involve lending from $ 1,000 to $ 10,000 per month (over a year) to use ANY resources from the IBM Cloud for your pilot projects. For more information contact us .