How often you have thought of learning to use machine learning algorithms or techniques by yourself ?
You always think of doing something in Machine Learning but you always get restricted by the constraint that how do you start, you do not how to program machine learning models and deploy them to see some real problem solving.
One solution could be the open source platform from IBM data science labs. A visit to the machine learning presentation at IBM research labs by Vishal Chahal, Lead Data Scientist at IBM tells you a different story at all. They have provided a platform called as IBM Data Science.
How about you just do not need to worry write any code at all and yet see the magic that machine learning is capable of? How about just building a data pipeline in the form of flow of data and machine learning model is built automatically for you? You can then use it to see output on real data.
All you need here is to provide your training data which you can always find it from Google.
For example, you want to do some image classification, you can build the machine learning model on this platform for this. You will also have the option to select models from a lot of available varieties. You will only need to provide images data for training the network for which you can write a python script to fetch data google automatically or you can download them manually and drag and drop the images into this platform, as simple as that. Your model will be ready in a flash which you can deploy within a minute onto IBM cloud which comes integrated in this platform only with a click of a button.
Machine learning in the modern world is finding its use cases in almost every sector possible wherever a decision making is required which needs to scale and where data set could be huge or could grow gradually.
A demo of NAO robot powered with capabilities of using machine learning model to make decision is a very interesting example. It can take input of a video stream and detect hand movements out of it, then it stores the data onto the cloud which then sends information to an IoT devices connected to a flight control and gives the command for its movement.
Once you register onto the IBM Data Science website and explore capabilities in machine learning, you will get to have handful of experience of seeing the various code patterns available for popular problems, and you can also run them with your input.
The essence is that you don’t need to be too much programmatically informed to start implementing ML models onto some data and see the outputs of it. Instead you can use already build in code patterns on IBM data science
and check performances of various ML modes available, further you can also build your ML model and train it onto this platform and deploy also to solve some real time problems.
One more important regarding ML deployment has been its scaling in real time scenarios is its scaling when concurrent requests come asking to give decision based on the input. IBM cloud services is pretty prepared for this at the moment and one of the best handling around 1500 request per second. It’s a perfect platform at the moment to explore ML from the scratch for a beginner.
You can explore IBM data science by visiting its website http://www.ibm.com/datascience
These were the thoughts after attending : Machine Learning and its connection to Robotics event