Machine learning is changing the way we do things, and it has started becoming main-stream very quickly. While many factors have contributed to this increase in machine learning, one reason is that it is becoming easier for developers to apply it. And, that is through open source frameworks.
If you want to learn about machine learning in a detailed way, there are some resources to guide through it. There are many frameworks available, but there are some which can help a beginner get started with.
Why we use Machine Learning?
Before we get too deep into this let's look at a simple use now we'll use examples of PayPal which uses machine leaning to detect fraud so that PayPal does is it uses this different machine learning tools and before started using machine learning a lot of people were getting by with a lot of scams and then once they figured it out they said hey we need a tool in here that going to be automated and easy to use and reliable and so they found the machine learning tools came in and then they're able to stop the different fraud.
Applications of Machine Learning:
In the application of machine learning we have search engine results, voice recognition, number plate recognition and dream reader if you think hard enough you'll find that this small sampling is just the beginning from automated cars to scientific discoveries any of these are all part of today's world of machine earning and let's take a look at the search engine the first one image if you were at Google or Bing or Yahoo.com and someone typed in a search to one of those and that goes across to somebody and they look at it they have to read it then they go to the library they look up the information they come back and they type back the answer to you which might or might not be correct depending on which person you got one of the things we notice right off the bat is that's not what happens nowadays you give very reliable information you get it fast , it's automated and as time goes on and we get more information the search engine returns better and better results this is true as a voice recognition where it gets better and better recognizing what we're saying and able to transcribe that for any of your Google commands or home devices where they recognize your voice or maybe it's Siri if it's iPhone we see that in number recognition we've seen that is being able to read things back to us like dream reader does with news things so as we're looking at applications of machine learning we want to take away that the use of machine learning or why use machine learning is because it makes our life easier and also helps to processes more consistent and reliable.
How does Machine Learning works:
This machine learning is broken up into three different steps:
Clean the data
The first step is you need to clean the data and format data that's your pre-processing which mean computers nowadays aren't too smart when it comes to figuring out the difference between a picture or text when you send it in so the first thing you do is usually clean the data so that all your pictures are in one file and your text is being processed separately and that's all part of the pre-processing if you try to do process text like you do a picture you're not going to get the right answer and vice-versa. Once you have processed the data and you have it nicely cleaned then learning take that data and learn from it and there's what they call supervised and unsupervised data. The third field is testing, once you have gone through all these processes now you have to trained your learning so it works correctly you need to test it making sure you get the right answers out of it once you have gone through all that then move to phase two which is actually using it or putting it into commercial use and that is to do a prediction and on there here you have to train the model.
Types of Machine Learning:
1. Machine Learning a Supervised Learning
In supervised learning, algorithms use training data and feedback from humans to learn the relationship of given inputs to a given output. The goal of supervised learning is to approximate the mapping function so well that it generates a new input data that can be used to predict the output variables. It is called supervised learning because the process of algorithm learning from the training data is similar to a teacher supervising the learning process.
Types of supervised learning:
Classification: Classification is concerned with building models that separate data into distinct classes.
Algorithms used: Decision tree and support vector machine.
Regression: Based on previous input data, the machine predicts continues output value.
Algorithms used: Linear regression and Polynomial regression.
2. Machine Learning as unsupervised Learning
Unsupervised learning is where the only input is data and there are no corresponding output variables. In other words, an algorithm explores input data without being given an explicit output variable an example is sales and marketers exploring customer demographic data to identify patterns of existing and potential customers.
The goal for unsupervised learning is to model the underlying structure or distribution in the data in order to learn more about the data. It's referred to as unsupervised learning because the answer is not known algorithms discover and present patterns and insights in the data.
3. Machine Learning as reinforcement Learning
In reinforcement learning the answer is not known, so the reinforcement learning agent still has to decide how to act to successfully perform its task. Because there is no training data available, the agent learns from experiences. It collects the training examples and through a trial and error process and it relentlessly pursues the goal of maximizing the long term reward.
4. Machine Learning as Deep Learning
Deep learning is a type of machine learning as
Process a wider range of data resources
Requires less data preprocessing by humans, and
Often produce more accurate results than traditional machine learning approaches.
In deep learning, interconnected layers of software based calculators can ingest vast amounts of input data. In the next step, the data is processed through multiple layers that learn increasingly complex features of the data at each layer. The network then makes a determination about the data, the accuracy of the determination and then incorporates what it has learned to make smarter determinations the next time. There is a direct correlation between time and intelligence.
Hence, the most recent advances in AI have been achieved by applying machine learning to very large, diverse data sets or data lakes. ML algorithms detect patterns and learn how to make models, predictions and recommendations by continuously processing vast amounts of data and experiences, as opposed to using a rigid set of commands programmed by a person or team of people based on the information available at a specific point in time. A huge differentiator is that the ML algorithms constantly learn and adapt as new data and experiences are processes and that means continuous learning over time plus they never tire and make far fewer mistakes than humans.