WSO2 Machine Learner: Machine Learning for Everyone

Machine learning is everywhere. If you do not agree, you have not noticed. It is everywhere. It’s inside your smartphone, web browser, smart watch if you use one, car very soon, and the list goes on making it simply everywhere. Machine learning has become one of the things which decides the future of mankind in one way or another. There was a time when artificial intelligence, the superset of machine learning, was something we read in science fiction. But not anymore.

Never heard of what I am talking? Do a simple Google search and you will end up with thousands of definitions, tutorials, articles and maybe some news on recent ML startup acquisitions. Google itself uses machine learning to bring all that to your web browser. So what is WSO2 Machine Learner? Why you should care?

image

WSO2 is one of the leading middleware companies in the world with a very unique business model. Let’s keep details aside, I’ll just say that they (or we) develop products that are free and open source. We offer a complete stack of middleware products which are used by top enterprises of the world for integration to API management to real-time analytics. WSO2 Machine Learner is the newest addition to the stack to make machine learning closer to everyone. When we say everyone we mean it. If you feel like trying out the product even before reading the rest of this post, go try it out! If you think you should look at the code and even make some changes to it, go fork it from Github!

Spaces: Community and tools

Glad you decided to continue. In this post I would like to take a step back from my developer role and look at our own product from a typical ML user’s perspective. You might be a researcher, developer or even a CEO who is interested in trying out machine learning in your next venture. Reading this will give you an idea why WSO2 Machine Learner can be one of the best choices you make when it comes to machine learning.

There are different communities in machine learning. Mainly there are researchers who make this field what it is and decide its shape in future, developers who apply the theoretical work done by latter in practical problems using dedicated tools and there are industry people who look at this field as a new opportunity to grow their businesses. CXOs might fall into the last category. Interestingly in machine learning, these entities usually overlap, making highly research oriented CEOs and scientists developing industry solutions. Most of the top scientists in machine learning (and deep learning recently) are leading the industry towards catering more futuristic products to end user with AI/ML. At the same time top machine learning startups have been initiated mainly by the best researchers of this field.

To make all these a reality, we need tools. Tools which let all these entities work more efficiently and effortlessly. If you look at that space, where ML tools are scattered, once again you will notice more distinct regions. There are products that are very easy to use and intuitive, but most of the time they are not free. And there are products that are very powerful and extensible, but you will have to get your hands dirty (dig into code and make changes). Most of the tools used by researchers fall into this category. No disrespect, they have their reasons use them (I had my own). Another set of tools will give you an amazing set of visualizations with a very user friendly UI, but not enterprise ready. If you are looking for all of the above today, you will not end up with many choices. We truly hope that this will change soon. And instead of just of hoping we thought of building a one, to deliver most of them today and more in future.

How WSO2 Machine Learner helps?

screencapture-10-100-7-51-9443-ml-site-preprocess-preprocess-jag-1442221876895

WSO2 Machine Learner workflow

WSO2 Machine Learner is powered by Apache Spark, popular and powerful large scale data processing engine. All the machine learning algorithms available in WSO2 ML are running Spark engine at the product core. With the recent boost from IBM we can safely say that Spark is here to stay, at least for a while. All the widely used classification, regression and clustering algorithms are available in WSO2 ML with a user interface that will make you think machine learning is easy! Yes, our web based interface will let you smoothly flow through the complete machine learning life-cycle.

  1. Get data
  2. Explore/analyse
  3. Build models
  4. Evaluate
  5. Predict/deploy

No matter from which entity you are from out of the entities I mentioned above, you will find it more convenient to do machine learning using WSO2 ML at no cost. The product will guide through each of the step, involves in this process in an intuitive manner and let you look at what you did with a comprehensive set of visualizations. Why visualizations? Because they talk in a language we all understand and that language is closer to us than numbers. These visualizations range from simple tables to sophisticated charts which will make data talk by themselves. At the end of the day your need can be either a low level task such as building the best ML model for a given dataset or a high level task such as evaluating machine learning as an opportunity for your business; either way WSO2 ML will help you to decide on what to do and how to do it. No matter whether you are a newbie or an expert it will let you get your job done with most powerful data processing tools available today. And when we say enterprise ready, we mean seamless integration with other products offered by WSO2 for integration to API management. WSO2 ML can be also thought of as a one component of a complete analytics platform that includes batch, real-time and predictive analytics. We do not discriminate, if you are just starting here and looking for something to begin with, simply get the binary and you are good to go! If you are looking for an enterprise level analytics solution with the power of machine learning take a look at our platform and decide which tools will work for you. In both cases, it’s free and it’s the same product.

ROC curve and AUC

ROC curve and AUC

Confusion matrix for Iris dataset

Confusion matrix for Iris dataset

Predicted vs. actual chart

Predicted vs. actual chart

What to expect?

Machine learning as a field is not going to end soon. Or I may say this is just the beginning. We are fully aware of the potential of it and the direction it is heading towards. In the initial release of WSO2 Machine Learner we are offering an essential but a complete set of features for your machine learning tasks. Alike to this field itself, our first release is just the beginning. We will be adding more and more features to WSO2 ML with cutting edge tools and research. In immediate releases you can expect multilayer architecture support, extensive data pre-processing, recommender solutions, anomaly detection and PMML support. And that’s not the end, we see beyond. Keep tabs on us, we will let you know.

Thanks for reading and happy machine learning! 🙂


Hate reading documentations but want to know more? You can just Sneak Peek into WSO2 Machine Learner 1.0.

Advertisements
Standard