Today we take another step toward making Machine Learning simple and accessible for everyone with our launch of IBM Watson Machine Learning.
Why are so few companies using Machine Learning today?
- Companies want Machine Learning, but they don't understand it.
- Requisite skills are hard to find.
- Machine Learning at scale is daunting.
- Competing technologies make for difficult choices.
- Deployment is very complex and not standardized.
Watson Machine Learning is designed to make AI and Machine Learning easy to use and understand. We want to support the full machine learning workflow and bring automation to every single step.
Our goal is to democratize Machine Learning (ML) and make it available to all types of users regardless of skill level. There are different user interfaces to build ML models:
- Model Builder: Not everyone is a data scientist, nor needs to know about model design, statistical theory, or training the model. Even developers, for example, may have varying levels of need when it comes to data science. They may just want to be able to use a known model that works well and deploy it in their app. We designed a simple flow that enables developers to not only choose and deploy a model but actually create a pipeline through a step-by-step process: Select data --> Train ML model --> Evaluate performance --> Deploy model --> Monitor.
- Jupyter Notebooks or RStudio: Create machine learning models by using Python, R and Scala. Feel free to choose your favorite libraries: Spark MLlib, Python Scikit-Learn, Spark ML Pipeline and many other. Then use the
model.deploy()command to get the web service API of that model.
- SPSS Modeler: Create a visual pipeline and build your models without programming. Use SPSS streams management and deployment with real-time scoring and batch-processing options.
Use your own data to create, train, and deploy self-learning models. Leverage an automated, collaborative workflow to drive intelligence into day-to-day business applications easily and with more confidence.
Select your data sources, review the data to ensure it's right for your needs, and prepare it for analysis. Select transformers and watch the pipeline being created as you work.
Train the model, validate the results, and check your model's performance. Take advantage of our easy-to-understand visualizations to evaluate your model and its performance. IBM Watson Machine Learning facilitates this type of validation by providing a clean separation of training data from holdout data used to evaluate model performance, as well as careful use of cross-validation techniques.
Deploy your model as an API, as a batch process, or into a real-time stream. Easily create apps powered by machine learning, or make existing processes smarter. A Machine Learning model becomes stale the minute you stop training it. Over time the accuracy of the models can worsen and can take significant time to understand what is happening. Our tools make it easy to retrain your existing models and deploy new versions.
IBM Watson Machine Learning is in a closed beta, join the waitlist and we will invite you in the coming weeks!
Learn more about how we are trying to bring Machine Learning to all enterprises in the article written by Dinesh Nirmal - Vice President of Big Data and Analytics at IBM: https://dineshnirmal.wordpress.com/2016/08/23/machine-learning-for-the-enterprise/