Machine Learning is changing fundamental of Computer Science and traditional Analytic. In Classical Programming we explicitly program a computer to provide some output for a given input. Its deterministic behavior.
However Machine Learning, there is no definite conditions or rules which determines the output for given set of input. In Machine Learning, like human being we train the computing model with given set of input & output which is training Data Set. After the training we validate the computing computing model against test data set. Once the training is successful, those computing Model has been used to apply Machine Learning on real input data set. The examples of Machines Learning. Machine Learning can be categorized in two types:
- Supervised Learning - This type learning model depends on input labelled data set. For example determining spam email. In this case Computing Model refers to predefined set of labelled Data Set to predict whether its Spam or not Spam.
- Unsupervised Learning - This type of learning algorithm applied on input data set without being labelled. It performs probabilistic regression analysis to predict the output. Example of unsupervised learning is link tracing or anti money laundering tracking where computing model does not have any label data set and determines relative patterns and predicts approximate output.
In our real world example apps google now is using Machine Learning to train its Apps understand each user voice pattern. So the more you use this app, the more accurate result it will provide. Its like true human being the more you practice, the better you perform.
There is huge potential to extend this Machine Learning Capability to Enterprise through SOA Business Activity Monitoring, Real-time Analytic. There are some tools like Oracle Stream Explorer, Apache Spark which can be used in Enterprise for Real-time Predictive Analytic to gain more insights quickly in the world of Social, Digital and IoT
No comments:
Post a Comment