MACHINE LEARNING
WHAT IS MACHINE LEARNING?
As we know that the future of machines is very huge and beyond our imagination. It is all around us in the modern world. Machine learning[ML] is basically in which humans train the machines to make decisions. In simple words, Machine Learning is " Machines that are trained from data ". It is the sub-core of Artificial Intelligence. ML Algorithm builds a model based on sample data known as 'training data'. It accesses the data and performs tasks automatically that can teach itself to grow and change when exposed to new data using iterative algorithms to make prediction and detection without being explicitly programmed to do so.
More data > Better Model > Higher Accuracy
We have numerous examples around us. Some example is as follows :
-If you say Alexa play the song. It will play the song according to the song you played the most.
-You must have watched NetFlix. There is a different section of recommended movies which you may like. It would recommend the movies according to your previously watched movies, web series, documentaries.
Therefore, machine learning helps in solving the highly complex nature of real -life-world problems. Thus, we need specialized algorithms that can solve difficult problems because manually it becomes very difficult to solve the problems.
TYPES OF MACHINE LEARNING :
- SUPERVISED LEARNING :
In supervised learning, the model can predict with the help of a labeled dataset ( i.e already known dataset) while training the algorithm. In simple words, the training dataset already knows the output the algorithm should provide. So a labeled dataset of mango would tell the model which photos were of mangoes. When showing a new image, the model compares it to the training examples to predict the correct label. It is further divided into two types :
- CLASSSIFICATION:
In a classification problem ask the algorithm to predict a discrete value such as yes/no or true /false. When a model is trained such that it has to answer with such values we make use of classification.For eg: the Model is trained such that it has to detect the images of cats. So, when we provide the data it detects the image and gives the output in the yes or no form.
- REGRESSION :
In regression problems the relationship b/w two or more variables is such that the change in one variable is associated with the change in another variable. for eg :
in linear algebra, the change in the value of x will also reflect on y.
- UNSUPERVISED LEARNING :
In unsupervised learning, the model is provided with the unlabelled dataset and the algorithm analysis on its own by detecting the size, image, etc. In simple words, the model is given a dataset without giving any desired output. It is further divided into the following parts :
- CLUSTERRING :
In clustering, the objects are divided into clusters that seem similar according to their shape, color, size as the images of birds are similar so it becomes one cluster and the images of dogs are categorized into another cluster.
- ASSOCIATATION:
- ASSOCIATATION:
In association, the likelihood of adding the items in a collection. Let’s try to understand it by example if you go to the market and buy burgers and french fries then there are chances for you to buy the coke as well. Thus, in an association, the algorithm is trained such that it provides with the items you are associated to buy.
- REINFORCEMENT LEARNING :
Reinforcement learning[RL] is a type of machine learning in which there are components such as agent, the environment in which agents traverse over the environment to train themselves. In simple words, when the agent performs certain actions if the action is appropriate it is rewarded with good outcomes and if the actions are inappropriate it is punished for the wrong outcomes. Therefore, it learns from its experiences after training the model by performing certain actions again and again and eventually it learns which action will lead to the best outcomes. Let’s understand it by taking an interesting and very trending example of today's lifestyle. Self-driving cars, if a car needs to take a turn but it moves straight so it will give the wrong outcome and it will learn from its experience and next time it won't repeat the same mistake on the other hand if it takes the turn so it will give the right outcome and reward it.
Very informative ,Keep it up.
ReplyDelete